Hacker News Reader

Comment by actionfromafar | Hacker News Reader

Martin Scorsese should sponsor https://github.com/oyvindln/vhs-decode :-)

It's the ultimate in VHS preservation.

I'm sure the Venn diagram between HN and Hollywood could make this happen!

I think the biggest obstacle right now is not the software, but the hardware. The software works, but requires either buying a $300-$500 set of three circuit boards, or finding a specific PCIe capture card and potentially modifying it. If someone made a nicer version of the Domesday Duplicator - single board in a case, $100 to $200 - I think that would make the project a lot more accessible. Even better would be if someone started selling modified VHS players, having replaced their normal electronics with a capture device and a USB3 port. After all, many people don't have a soldering iron or a desktop computer.

If we are talking about serious archiving of lots of tapes, $500 for a capture device is nothing. The labor costs of keeping the actual tape deck(s) in good order is probably going to dwarf.

But I have actually toyed with the idea of setting up complete workstations modified player, and a slightly better software UX. Don't know how to reach the market willing to pay for such things, though.

For tinkering at home, the specific PCIe capture device is actually quite easy to find, it seems it's still produced for some reason.

I’m sorry what is this? Is this a software capture card? I’m so confused.

It appears to be software designed to use an SDR or (and?) a DTV capture card to capture the output of a VCR playing a VHS at the RF level, and then transcode and store it. Pretty cool, if I'm understanding it correctly!

How lossy is transcoding at the RF level of an already lossy medium?

It's much better than any other existing technology to capture VHS.

It bypasses all analog processing of the signal by the VHS player and captures the video signal directly from the tape head.

All further processing is then done in software. If any improvements in software processing is made, the "raw" capture can then be reprocessed with better results.

So it's like a Kryoflux for VHS tapes? That is cool.

Ha, adding that to my parlance! It's exactly like Kryoflux for VHS tapes.

An example - S-VHS tapes can not be played back in regular VHS players. They can't handle the format.

But with vhs-decode, it doesn't matter. A regular cheap VHS player can "rip" an S-VHS tape.

And decode the HiFi FM-stereo signal, too.

Or Domesday Duplicator for LaserDisc Preservation.

It’s actually a fork of that project!

Best at capture/preservation, but only theoretically best at playback someday. Currently the best software processing algorithms are imo not on par with video playback quality from the best hardware of the era, especially for damaged tapes. It's hard, although certainly not impossible, to imagine that changing. I'm 100% glad it exists though.

You could always take the best hardware of the era, and capture from that. If you were talking about decks. I'm not sure about playback - the output from vhs-decode looks pretty darn good to me, but I'm not an expert at all in composite video gear.

Even the best VHS decks from the 80s/90s, which would have been used in production environments and thus “well worn”, would have wear-related quality issues.

But that’s the beauty of vhs-decode. The transport mechanism needs to be stable and the head clean. The rest of the player doesn’t matter.

My impression was that the transport mechanism would have seen the most wear.

Right, my language was sloppy but I meant two things-- the quality will still be dependent on the hardware chain, and I should expect that a 'bit perfect' capture of a VHS tape will usually be inferior to one captured from a vintage pro deck. And then regarding processing, I'm not a professional, but I've done several successful restorations using virtualdub and avisynth, and on some degraded tapes I couldn't do nearly as well as whatever magic was going on in the vhs to dvd section of the JVC SR-MV50.

I'm not following why would software still be inferior playback to now ancient hardware? Isn't that sort of like suggesting digital can't replay a vinyl record transcoded to digital as well?

Sort of correct - except that the software to reliably decode VHS took some time to get right, and there still are regular improvements to the vhs-decode github repo.

For a long time a great S-VHS deck, professional Time-Base-Corrector and so on from a TV studio, was the better option over vhs-decode software, especially on bad tapes. But now, I'd go with vhs-decode any day.

No, I'm thinking of proprietary hardware or software in the decks that perform tbc, noise reduction, filtering of artifacts, etc. If given a choice of a perfect copy of a spotty VHS and all the virtualdub and avisynth filters, or the mpeg2 output from a JVC SR-MV50, my experience has been that the latter is better. However, I grant that a video preservation professional might get better results from treating a direct capture than I.

Yes, but still don't throw away the original tapes, because if a better (read: more accurate/sensitive) tape head ever becomes available, you'll want to use it to upgrade your capture device and then re-scan the original tapes with even better fidelity.

It's a fully software-based VHS decoder. It takes the raw signal from a VHS tape, captured with a high speed sampling device, and reconstructs the video on your computer. In theory, this approach should allow higher quality digitization than any physical VHS player - it cuts the analog decoding electronics out of the loop, and it allows you to capture once, then tweak the algorithm and decode many times without disturbing the tape. You could even preserve the raw captures in case better decoding algorithms come out in the future.

Reply on news.ycombinator.com