Been quiet on this for a while, but I just rewrote a good chunk of the TBC and that bit's about 8-10x faster. Once the comb filter is optimized, the CUDA version will probably work in real time

Awesome. I've been distracted for a while.

Are you doing anything special to decode the superntsc discs? When I play a superntsc disc through my superntsc decoder, the image is pretty amazing. There is essentially "zero" comb filter issues, and the color resolution looks really good too. I suppose it isn;t really cost effective, so few discs are encoded that way...

I bought a CFD-N that came with a complete schematic, but I haven't been able to find any good literature on what the encoders/decoders actually do, and I don't know enough EE to decode the schematic.

Tony

Not yet, but I might take another look when I get to the comb filter in the coming days/weeks/etc. There are quite a few references on it - look for an older edition of Video Demystified, there are .pdf's floating around on the net. If you have access to a university library's journals, I think there are Faroujda papers on IEEE xplore.

Faroujda didn't believe in 3D comb filtering, so a SuperNTSC decoder would (primarily) be an enhanced 2D comb filter.

And many/most later disks are SuperNTSC encoded, but the biggest downside is a lot of those disks also have early DNR. By that point LD (pre-?)masters were digital.

(sidenote: Even if I don't have anything worth mentioning here, when I do work on ld-decode I regularly post code/Python notebooks on github)

Hello,

I am quite a newb when it comes to all electronic stuff, but I would love to try out this great software you have made here guys. I have same player that Megapixie has - LD-V4300D - could you tell me all the stuff I need to purchase in order to get this working and running? Do I need amplifier for the signal? I am not quite satisfied with video coming out of my composite out on the player, so I want to get the best out of it.

A 4300D in the US? Cool.

The best capture card I've found is the WinTV 34132, but it's pretty expensive (~$50 on eBay right now) so I'd look for alternative cards using the CX2388x chip, preferably without the MPEG2 bits. (sooo many Hauppauge MPEG2-encoding cards on the bay right now, but I doubt they can do raw mode)

You don't need an amplifier in practice I've found. Recently I've had good success with BNC cabling, it has less RF noise. This is a nice s-video->BNC adapter: http://www.ebay.com/itm/141786425357

Some pictures (including ones from megapixie) are here: https://docs.google.com/document/d/1n5V ... OYoRg/edit

(general update: Largely debugged my fast TBC code, going to slowly? merge the CUDA and non-CUDA versions of the RF demod code. Then hopefully onto the comb filter.)

Thank you guys for quick reply, I am not from USA actually, but from Poland I dunno why this happened on the forums, I got automatically assigned that location. I have purchased PAL and NTSC WinTV cards to try out, bought the cable you're mentioned. Now I just gotta wait few weeks to delivery (importing from USA takes some time ). I am extremely hyped to see the results with this.

Cool And oops, I should've mentioned the same cx2388x card can be used for NTSC or PAL. That said, every card design has different noise properties, so it's worth trying a few to see what works best.

I haven't done much on the PAL side, but the early results were very promising. It really needs digital audio decoding to be useful though, since I can't fallback to the analog tracks on later disks.

I hate to kick up an old topic; but I'm wondering what the status of this project is and how much bandwidth is needed on the ADC end to accomplish it?

Planning on working on it again over the next couple of weeks - in particular a faster comb filter. It probably needs a ~25mhz 8-bit ADC for good quality, but it can probably be fudged slightly.

After reading through this, I remembered a piece of hardware I came across the other day:

http://store.newae.com/chipwhisperer-li ... sic-board/

It's meant for something completely different, but it should be generic enough to do exactly what is needed here?

Another thing I came across today, that I'm guessing everyone here has already seen... But just in case:

http://www.kallhovde.com/pioneer/tricom ... RV2125.pdf

That should be a good place to start, if someone feels like figuring out how to demodulate the AC-3 audio.

Anxiously watching for further updates in this thread!

That board could probably be made to work with some effort, depending on how fast the USB channel is. Right now I'm still using the CX23880-3 chip, I got an Airspy but it doesn't have quite enough bandwidth.

I've seen that service manual before, what I need for AC3 decoding is more bitstream info - that's all in the main chip, so it's not documented in the manual alas.

If bandwidth/bits are an issue, would something like this work? (With a decent sound card)

http://fivedash.com/index.php?main_page ... ducts_id=8

Or, would that introduce too much noise? Not work for some other reason?

A sound card solution wouldn't even have enough bandwidth to decode the audio channels... and in the case of AC3, the audio's signals there but I don't know how to write anything to decode it yet.

Oh, I see my mistake. I was thinking about a 24 bit / 96kHz sound card, but 96kHz isn't even close to 28MHz.

I've been contemplating buying one of these:
http://www.digikey.com/product-search/en?keywords=AD9265-80EBZ-ND
($217 eval board for an Analog Devices 16-bit ADC -- 79 dB SNR at 70 MHz! )

and interfacing it to a PC using one of these:
http://www.digikey.com/product-search/en?keywords=428-3347-ND
($46 eval board for a Cypress programmable USB 3.0 interface chip)

I *think* the Cypress interface chip can accept a raw bitstream, but if some sort of framing/buffering is required, an FPGA board could be inserted between the two. This starts to get pretty complicated, though.

I'm a little reluctant to drop 300 bucks, plus whatever it takes to make an adapter board that connects the two together. I have a sneaking suspicion that the results won't be significantly better than what the WinTV cards produce. Still, it would be nice to know for sure.

Yeah - and it would require extra software effort too. Might as well stick with the WinTV for now... if something like that comes down to under $100 or so complete, I'll get one m'self!

Wow, I am very interested in this project. I have a bunch of LDs that I would like to capture in the best way possible, but I have not yet bought a player. I was originally thinking of getting a Panasonic LX-900 (if I could find one), but now I'm considering the Pioneer CLD-D704 because I believe it should have the RF test point.

Anyway, apart from the player, there's one big problem for me: I don't currently have a modern PC with PCI slots like these WinTV cards need. My main PC has only PCI Express, which means I would have to get or build another PC to do this.

I've been thinking about getting another PC anyway in order to run Windows XP and software from that era, but it sounds like from this thread that the system would need to be pretty beefy to handle capturing the RF. What do you think I would roughly need to make this work?

It's good to have a dedicated machine for this sort of thing -- no surprises/interruptions/slowdowns while you are capturing. I'm using a Pentium-4 2.7 GHz PC with a clean 200 GB ATA drive to store the captures. It seems to keep up just fine. I bought an SSD to use for this, but I haven't found a SATA adapter that works in this PC at speeds higher than ATA-2 (33 MB/sec), which isn't fast enough.

Anyway, you don't need a super-powerful PC, which is good, since finding one with a PCI slot can be a problem...

Thanks for the response! It sounds like I shouldn't have a problem getting a PC to do this then.

I do have another question: would a capture card like this one work? It uses the required CX2388x chipset, but also has a second Conexant chip with an MPEG-2 encoder.
http://www.ebay.com/itm/221589033404

Not sure. I bought a random one off of ebay and got lucky, but it doesn't have the MPEG chip. I seem to recall that the cards *without* tis chip are more likely to work, but I don't have the source reference handy.
One interesting thing about that card is that it has a composite input jack. Depending on how the inputs are wired, you *might* not need a hacked s-video cable to get RF into it, or a little minor surgery could re-route the required input to the composite jack. It's not critical, but would make wiring easier.