Thanks!

I'll post a longer video (I'm thinking the rest of that rock adventure side, someone on youtube asked me about it and it hasn't been sold by anyone in years) once I have CX decoding figured out.

AC3 is tricky not because of the encoding (which is well documented in the patent) but because I have no idea what the actual error correction specs are, and I don't know who to talk to. (Disclord was probably the best person to reach them through. ) It would be nice to do, especially with AC3 decoders becoming quite expensive on eBay.

Also, always test with a rotter/soft picture disk when trying to align a player. (I thought I tuned the tracking of one of my 2800's until I tried to play my Discovison-pressed copy of Airplane! )

Nice stuff!

Thanks! Here's the entire first side of Rock Adventure, with a basic 3D comb filtering and better (but not CX decoded yet) audio:

https://www.youtube.com/watch?v=tqoYNbh ... e=youtu.be

And now off to camping.

This looks pretty good. I'm glad to see that you are still developing this.

I had this thought of running a fiber optic cable from the pickup head down to a toslink input on a soundcard and getting the information that way. I'm sure there's a reason why it couldn't happen but it sounds cool so I thought I'd mention it.

Happycube - looking awesome !

The 3D comb filtering is looking equal to what I see from the CLD-99 and Snell and Wilcox TBS-185. The only room for improvement is dot crawl on the color bursts bottom center.

My recent Sony TV is doing something to suppress the dot-crawl in that case, but I have no idea what.

Good stuff.

Seems very good! May you post "A Video Standard" and/or "Video Essentials"?

(I came back from camping with a cold, hence the delay here. Ah well, getting better now!)

robwhar: A sound card's TOSlink is meant for SPDIF signals - the # of pits on an LD is an order of magnitude or so beyond that, and getting their timing is essential. The way laser pickups work, at least in LD, gives you an analog signal at the end. The pits are basically pulse code modulation for an analog FM signal. A CPLD (simple programmable logic chip) might be fast enough to record and process pit timings, but I didn't see any need to go that route.

I now have a pair of LPCLink boards, and I'll see what I can do to get ADC streaming over the USB port, then I'll try to hook it up to my LD player. If that works it'll be a cheap ($20+cables) USB general ADC device which will have other uses, no doubt!

Semi-related - it wouldn't be hard to do standard digital audio decoding, that just hasn't been a priority since there are already bit-perfect ways of preserving it. But it could be useful for people who want to do verifiably perfect CD rips...

megapixie: Thanks! Your comb filter post (and other data on it) has been very useful for comparisons. I'll have to remember to get a longish capture of something w/closed captions, then I'll send you the CC line dump as a thank you. My wild guess is that the Sony is adding/averaging frames, or something along those lines.

laserdisc.ws: I've got the entire montage of images from VE captured, so I could post the video and left channel analog audio. I have AVS but haven't captured any of it yet. Also not sure what to do about interlacing on those, since there are mixed film and video segments.

Posted the second side of Rock Adventure, with CX decoding now (partially?) implemented. I've never played with audio compaction/expansion before, so it's still probably a bit off but sounds pretty decent, aside from rot/speckling related crackling. Anyone got good CX test signals? The one on the reference disk isn't really suitable for getting the timing envelopes right.

Now it's time to make it all comprehensible...

If you are able to highjack the raw signal and process it to produce an image quality above that of your typical player would there be the future possibility of an FPGA mod board that could be integrated into an existing player and provide an HDMI output? I understand that is a fair jump from were you are now but it would certainly be cool if it could be implemented.

It's theoretically possible, I just don't have the time to make one. It would probably involve an original circuit board to handle the ADC input and HDMI output.

I have been considering this for sometime. I don't think the digital to analog stage on ld players is the bottleneck. It happens in 10bit and fairly simple artifact free process.

For hdmi/dvi or sdi you need 27mhz component signal coming from the video board. This means the player will use its own comb filter to separate and you need to implement a y/c to component conversion before you can feed to a hdmi transmitter. Another issue is ld is 14.3mhz and in component it would be 28.6mhz. You need to convert this to 27mhz.

You can do composite digital sdi but only few broadcast grade device would work with this signal.

For hdmi you need to build the entire digital section from scratch for it to be worth it. Tap to analog signal, use your own 10 or 12 bit analog to digital converter then tbc comb filter and so on.

It would make more sense on a machine which has a good analog section with clean steady signal.

Yup, the digital TBC is pretty much transparent. On many players it's 8bits+sync, oversampled ~3-4x depending on if you're looking at luma or chroma.

The main sources of quality loss are in the demodulator stages (deemphasis in particular) and Y/C recombination errors when the comb filter gets it wrong.

edit: As for the project itself, I'm still plugging along on it. Simplified the TBC a bit, and I'm looking at the 2D comb filtering some more since it causes hanging dots. There are definitely some better techniques I'm not using for it!

I want to give this a try myself. I'm planning on tapping the RF on the audio board on my HF9G and then disconnecting composite output #2 from the inside and hooking up the RF to it and then I'll just use a composite cable and run it to a wintv 34132 card and use a Female RCA->male F-type plug adapter to plug in to the card. Does this sound about right?

I haven't figured out how to use all of the stuff once I get it all hooked up but I wanted to make sure that this is the correct stuff I need to get started.

That's the right card. You also need to load in my special raw capture driver - https://github.com/happycube/cxadc-linux3. Set that up first, and then capture regular composite signals and running them through the TBC and comb filter programs. (I'll make this easier to set up over the coming days, I've just finished rewriting a huge chunk of the TBC code over the last few days/weeks and I need to get composite input working with it again.)

(Also, this goes in via the S-Video port, not the tuner port. If you're lucky you got the S-Video adapter with the board, otherwise you'll want to rig up a composite->S-video cable. The standard adapters don't work since they have filters, IIRC)

You need to tap the full-signal RF, which is brought to the test point strip on the main board. The audio RF may already be low-pass filtered basically destroying the video signal.

A signal amplifier is also a really good idea. I use a VGA amplifier with adjustable gain - something like the Ocean Matrix/Kramer OMX-4000 is pretty good.

And if you have one, mod a cheaper player, at the same adjustment quality a CLD-D503 would virtually be the same as the HF9G/CLD99 for instance. I'd use a CLD-1xxx player there - if you capture me samples from PAL disks, I might even add PAL support someday

Thanks for the information. That should be enough for me to get set up . I haven't ordered the wintv card yet so I'll look for one with an s-video port. The one I'm looking at on E-bay has 1 actually so it shouldn't be a problem. I'm not sure how to rig up a composite->s-video cable without using an adapter. Would I just run the RF signal into the Y pin or something? It might be better for me to cannibalize one of the s-video ports on the player then instead of one of the composite outputs.

I'll be tapping the RF at the point that it enters the audio board so as far as I can tell it's not been filtered yet. Once I get the top off it might be easier to find a point earlier in the signal flow. I sold my cld-d505 a few weeks ago so the hf9g is the cheapest unit I have now lol. I'll be careful though and should it not work, the soldering job should be totally reversible.

New sample videos up:

https://www.youtube.com/watch?v=ONxAZsKcZYM - revised Rock Adventure side 1 using current code. Better TBC and color handling. A little bit of dot crawl in places, but you have to look for it

https://www.youtube.com/watch?v=cbHsRggwYq0 - S&W zone plate - still not handling moving well, but quite well otherwise.

Wow, it looks great! The 3D comb filter looks excellent as well.

Really stunning! You bring laserdisc one generation up! It looks like a master!!!

Thanks!

Uploaded a Pioneer Reference Disk (HE-010) which is a much better quality disk with BGV-type clips, and some test signals at the end: https://www.youtube.com/watch?v=OBnO-uK7XKc

Some stills - ffmpeg does an excellent job of chroma NR without asking, so these are a bit noisier: http://imgur.com/a/0TcE7

I fixed A/V sync, and cleaned up the 2D comb filter a bit more, but it can still be fooled - some very advanced techniques (Faroujda, neural networks) are said to minimize this, but I played the disk on my D406+Panasonic TV and it rainbow'd in mostly the same places, which makes me feel a bit better about it. The exact code used for the youtube clip is tagged as v0.6.

Are you joking? Its THIS GOOD already? This looks AMAZING.