Wow, Only just found this.

I have a few very rare discs that I would love to capture in this way and am considering buying the hardware right now.

The first disc that I think would be good for testing PAL is the British Garden Birds by the BBC as it has Teletext digital data that could be checked quite easily against the reference engineering pages on the disc. The covers were not printed for this disc with the teletext logo until it was understood if the laserdisc format could support the frequency rate of teletext reliably in a selection of players.

Three of the engineering pages from the above disc.
https://picasaweb.google.com/114059900192747179235/LDarchive#6155143832851381154
https://picasaweb.google.com/114059900192747179235/LDarchive#6155143822030526018
https://picasaweb.google.com/114059900192747179235/LDarchive#6155143822635646578

I have more if needed and if i can get the right equipment will capture the disc for you if it will help!

I also have a PAL disc that has the test cards on but I cant remember which one it is from the to of my head.

But my real aim is to capture the BBC Domesday Discs which is described in much more detail here:
http://www.atsf.co.uk/dottext/domesday.html
(By one of the original producers who also worked closely with Philips on the tech spec)

Only problem is that they are CAV - Non Standard PAL with with a different color range with analogue audio or digital data depending on what portion of the disc you are looking at. (Audio for the video portion of the disc and data for the stills.)

There are only about 15 working players in the world capable of playing these discs and I am lucky enough to have a working player after 10 years of trying to fix it, I also have the full catalog of the AIV (Advanced Interactive Video) discs.

The reason for looking at this is that I have managed to get a 1:1 copy of the digital data tracks by accessing it via the SCSI interface but was trying to work out the best way to get the video as the Player has RGB-S output and ideally wanted to go back through the video path as far as possible for the best signal (Preferably before the Genlock - Explained later).

The player is so unique that it has an RGB input for an external video source and it will match the rotation speed of the disc so that video mixing can be done in analogue electronics without a digital frame store. I have manged to get the rotation speed to be dropped to about 20fps by supplying an external sync.

The player is almost 30 years old now and I know it will not last forever. Although the BBC have recovered two of the discs I would like to recover the remaining 3 titles. If I can get this to work then I should be able to recover the data with a standard player as the data dump of the first disc took three weeks as it needed to go a frame at a time and dump 350MB (Per side) over a 9600 baud serial connection from the host computer. (It was the only computer that could talk to the player and only had 128K ram and a 4Mhz CPU from 1985)

The discs that I have a 1:1 copy of the digital data tracks just happen to be the discs that the BBC have recovered most of the stills from the master tapes so there is also a possibility of of using them as reference material for correcting the color space.

Need to go and look at my bank balance.... then again....
(I have a spare 2 processor quad core xeon rack mount server lying around so half way there, just need some fast discs! and the right capture card)

Also if i can capture an NTSC disc with AC3 I also have a few AC3-Demodulators here so could potentially capture the demodulated AC3 output from optical SPDIF and send you the datastream and then possibly post you the disc it came from!


Thanks for all of the hard work!

That all sounds awesome! Is there existing software to decode Teletext?

And I didn't realize the BBC had only preserved two of the five Domesday(-type) disks. I'd definitely be interested in helping get the other three saved!

I don't have digital sound decoding yet, but if the disk can play all the way through in a standard player with digital output that transmits subcodes/non-audio data, you can probably recover the data that way. If that doesn't work the Pioneer LD-V4300D has an EFM out port that a capture system could be rigged to work with, I would be able to deal with that bitstream a bit sooner.

A reasonable size disk is fast enough if you're not hitting it with anything else at the time.

http://www.ebay.co.uk/itm/KWorld-X-Pert ... 4aea199cad - I have another one of this card so I can probably figure out how to wire something up to the odd connector and get the card working. 20 quid w/international shipping.

For teletext, i use EyeTV on my Mac, and i have a soft (on the Raspberry Pi) to generate Teletext.

But it's not complicated, it's just "data" on the image, in the v-blank interval.

Ive just found this for decoding teletext from VHS.

https://github.com/ali1234/vhs-teletext

Might help.

When I get paid I will have to start playing with this.....My Plan of action...

Sample a PAL disc with teletext and analogue audio.
Sample a PAL disc that has the BBC Test Card on one of the frames
Sample a PAL disc with digital audio
Will also capture the Optical PCM output
Sample NTSC Disc with AC3
Will capture the Optical PCM output
And capture the AC3-RF output
And capture the Optical out from the AC3-RF demodulator

Ian

Sweet - and that's an interesting technique to get teletext out of VHS.

That sounds good - although the (non-AC3) digital audio optical out is a bit far removed from the raw signal to be really useful except for verification. There's a lot of descrambling going on between the two, although it could be useful to capture the amplified EFM data going into the CD audio decoder if the sync was perfect.

The AC3 RF out is untreated (audio) RF, so it's included in a full-range capture.

Haven't been up to much on my end, but I should have a basic RF amp circuit this weekend that may or may not wind up being useful. And I'll probably start the 'final' comb filter code soon, focusing on adaptive 2D PAL and NTSC at first, then 3D hopefully.

Also been working on EFM/CD decoding a bit, I have about a 50-60% frame success rate decoding CD RF, but none yet with the squeezed/altered LD digital sound form.

Hello.
I have gotten interested in ld_decode, and have some questions.

1. I hacked up a video capture card to work as a direct ADC, then capture the raw RF signal from a test point on the Laserdisc player. This bypasses the signal processing/video output electronics, while the player itself works normally.

The raw signal is then demodulated into (analog) audio and video, then framing and comb filtering/color decoding are done in software. http://imgur.com/a/7VNff shows each stage of the process.

2-3. I think the X0 would win in either case, at least until the analog->digital conversion is upgraded. This is the best direct comparison I could come up with between the X0 and ld-decode (using a CLD-V2800, a rather low end player!): http://screenshotcomparison.com/comparison/128845

If my tinkering with an RF amp circuit is successful, the gap might close a little bit. Eventually I may cook up a circuit using the LPCLink2 (with the same LPC4370 chip found in the AirSpy), which I suspect will give better results.

4. Not much at all, I need good quality VHS RF samples (and/or the time to hack up a VHS deck, I've got a couple of Panasonics to butcher sooner or later). Also note that dealing with Macrovision will not be a priority.

I know you already stated that you haven't done much with vhs, but will you also support other analog video formats, like s-vhs, w-vhs, video8, hi8, ced and the various analog betamax/betacam formats?

If someone sends me good RF samples, yeah. Otherwise it's rather unlikely anytime soon, I haven't even done any VHS captures yet...

CED would be closest to LD out of those, it uses similar encodings (with a much lower color frequency, and CX-20 instead of 14). Once I have VHS other tape formats should be easy, to do VHS properly I'd have to have color under code and probably a very different TBC program. And don't forget U-Matic, there might still be old master tapes floating around that could use capturing.

Right now I'm trying to get CD EFM decoding working. I have about a 95% success rate on the frames it actually bothers to see. I'm going to have to rework some of Sidney Cabot's code to find the others I bet.

Would this card still be available....



What lead needs to be made from the player to the capture card. (Does this attatch to the standard Composite video in on the card or does it need to go direct to one of the pins on the cx23883 IC to remove filtering.)

Im guessing this needs to be something like RG174 and a short length to reduce losses and noise.

If there a better ADC available with better 10/12/15 bit samples with higher sample rate for a sensible cost, would you have any suggestions and would it work with existing software

I even have a DVL919 that has lost the left audio channel that I can use for some testing.

Thanks again

Here's a good one. Just wire up the RF feed into the composite port. With late players it even does fine without an amp (but a good one would probably help), you can go straight to the test point.

http://www.ebay.co.uk/itm/Hauppauge-Win ... 3f2d7fd934

I haven't explored other ADC options yet. I think something based on the LPC4370 (LPClink 2, Airspy) would hold promise, but there isn't anything that uses it which is quite right without some tinkering.

I was digging through some old documentation about the Philips VP-415 BBC Domesday Project and remembered that Richard Russell (Ex BBC) had something to do with the design of the video circuitry / PAL standard in the player. (Specifically a modified PAL that can deal with freeze frame on two fields 'Odd / Even interlace' - Something to do with PAL needing 5 fields for the carrier frequencies to get back into sync after a full cycle i think)

I then managed to meet Richard at the BBC Micro 30th Birthday Bash at ARM Ltd Global Headquarters where I actually had a conversation about software decoding of PAL for recovering the other Domesday Laserdiscs. (Where I also met Eben Upton and used a Raspberry Pi before they were out on public release)

During the conversation he mentioned the BBC transform PAL decoder from BBC R&D...

Thought you may like some of the technical details on how the decoding works.

http://www.jim-easterbrook.me.uk/pal/
http://www.bbceng.info/Designs/designs_ ... trbbc.html

Ive just got the WinTV Card...

Is there any Preference on distro and version of linux for the tasks.

Will I be able to capture on one machine and crunch on another as the WinTV card wont fit into my rack servers
(1U and 2U low profile) but they have LOTS of crunching power (6 quad core xeons between them)

What Spec is needed for the box doing a capture
Will I need more than 4GB RAM
How much disc would be used for 2 sides of a laserdisc (2 Hours)
Do you think it possible to capture across a network (I Typically got 80-90MBps to my NAS over Gigabit)

I know you have said that the processing phase is not real time, but im wondering how long it would take to process an hour of video. Is it worth me processing two sides of the same disc on different machines?

Thanks again and im hoping to have time this weekend to try and set up my first capture.

Cool. 4GB RAM is plenty I'm sure. Assuming you're using buffering there won't be any problems sending over GigE, it 'only' needs 28.7MB/sec sustained. Intel chipsets recommended, one of robwhar's older AMD box couldn't do the capture (but a quad core does better).

About 100GB/hour is required. I don't support PAL audio yet, analog would be relatively easy to add (can probably have it next weekend if not sooner, just got home from camping and I'm wiped ) but digital is a long way off (if ever at this rate) - for standard capture setups I'd recommend getting a bit perfect digital cap out of the player. Using two machines is good, and if you set up the scripts you can do both sides on the same computer with some speedup (probably close to 2x if those are 2-socket systems)

PAL 2D CF is still not adaptive, I'll see what I can do about that soon. It'd be awesome if you could take the PAL TBC output and feed it into the transform decoder.

I'm running Ubuntu 14.04 myself. Other recent enough distros should be fine. Go here for a breakdown: https://docs.google.com/document/d/1n5V ... sp=sharing - and if you have a google account send me your email and I can give you write access.

I think 6-10 hours depending on which quad core Xeon it is.

No updates on digital sound, but I've been playing with PyCUDA on a GeForce 740 4GB GDDR5 to get the hang of GPGPU programming. Video demodulation is now at ~3x real time (same quality), and my WIP audio code about 1x (with some quality loss)

So if/when audio, TBC, and comb filtering are fully sped up, and one has a faster video card (the 740 has "only" ~800GFLOPS) running this in real time might actually happen.

Hello, please excuse me if this is a stupid question, but I was just curious if it is possible to tap the signal before it is converted to RF?

Like the actual PWM pulses of the laser? That could probably be easily captured by simply polling at a static (but really high) frequency and assigning 1 or 0 based on that. You could then have a binary stream that represents the pits and lands.

I realize this would be more work, because you would then have to construct the RF signal from the digital data, but then you could ensure there was no degradation from the conversion process in the LD player and no loss of quality or interference from the cable connecting it to the capture card.

Tony

What comes off the photodiodes in the player is basically the RF signal. The pickup is not so precise that it can precisely tell where the pits are.

Dig up Pioneer Tuning Fork Volume 6 (on hifiengine and audiokarma forum) for more info. It's got a good description of LD as it existed in ~1982 (the LD-660/LD-1100/PR8210 generation of players) - 7 and 8 cover CD and LD digital sound.

Wow, this is fantastic work! I'd thought about doing raw RF capture in the past, but I dismissed it as impractical. Practical or not, you've managed to do it, and that's pretty awesome.

I've worked for many years on capturing video from laserdisc-based arcade games, for archival and emulation purposes. Software decoding of raw RF captures has a number of potential advantages, so I am excited to see how this develops, and I hope to find a way to help.

I've set up a system with a WinTV card and your software toolchain, and successfully captured and decoded a few test clips. Results haven't been stellar so far, mostly due to player condition and/or insufficient RF signal level to use the full range of the ADC.

CLD-V2600 - best overall, but noisy and jittery
CLD-1010 - had significant crosstalk; needs adjustment
Panasonic AG-LD20 - RF level was way too low

A few other interesting players I hope to try:

PR-8210 (gas tube laser)
LDP-3600D (NTSC/PAL)
HLD-X0 (not currently working, sadly)


I'm also planning to build a buffer/preamp circuit for impedance matching and better control over signal level for the ADC. If I come up with something that works well, I'll be sure to let you know.


If it would help with your digital audio decoding efforts, I could capture the SPDIF output from a disc with an LPCM soundtrack, so you'd have a reference to use as a target. (Or have you already done this by now?) I have an AC3 RF demodulator here, so I may be able to capture that bitstream as well.


Anyway, nice work, and good luck with your continued efforts!

Cool! And thanks

Look at using live-fft ( cat /dev/cxadc | ./live-fft.py mega) to make sure there isn't too much noise from the cables - it's in the google doc.

Most of the testing has been done with later players - either the last generation (>=1995 CLDs, anything based on the 504, or the v2800), and the slightly earlier 503/703-type mech has been observed to have cleaner captures. I'd be interested in seeing 1010 samples once it's aligned, and gas tube and X0 captures would be nifty!

A good amp/EQ circuit would be very very nice, I don't actually have much analog circuit experience so I haven't pulled one off yet.

The TBC can correct for the wavyness in your sample pic easily as long as the demod quality is half-decent. I'll check in a script I've been using that does all the decoding phases tonight, I can't believe I haven't checked it in yet except for PAL (oops!) but the encode scripts show how all the programs link together. (edit 10/31: finally just did.)

Some help with PCM and AC3 would be teriffic too, I haven't looked at either in a while and having some RF and decoded data could be very useful. Thanks!