It is currently 12 Dec 2018, 08:31




 Page 1 of 22 [ 428 posts ]  Go to page 1, 2, 3, 4, 5 ... 22  Next
Author Message
 Post subject: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 30 Apr 2013, 06:37 
Hardcore fan
Hardcore fan
User avatar

Joined: 18 Apr 2012, 18:02
Posts: 1468
Location: United States
Has thanked: 26 times
Been thanked: 19 times
(more edits coming: Go to page 10 for more current info and sample links, it's improved a lot since these early posts!)

--- original post

I've been working on decoding Laserdisc video directly from the RF signal, bypassing most of the player electronics and hopefully (eventually) getting cleaner pictures.

I've been using an LD-V8000 with the RF test point hooked up to a video capture card using cxadc for raw 8-bit 28mhz capture (which I updated for Linux 3.5) to get the RF signal, and wrote some very slow, dirty, code to decode the laserdisc image, and another program to do time base correction and NTSC color decoding (which amazingly hasn't actually been done by anyone yet, as far as I can tell... I was rather surprised by this)

I've posted the rough source to github: https://github.com/happycube/ld-decode - I did a little bit more work on this right now, before I head off for my real job ;)

There's a .jpg there with color bars with correct colors, albeit fuzzy.

The next steps are to clean up the software decoding (I'm thinking locking onto the analog audio signals, subtracting them from the RF, and then working out some sort of proper PLL for the video rf), combining the NTSC decoding phase using more information from the LD decoding, actually properly adjusting my players, and an AC3 mod but taken from the core RF signal.

(and/or purchasing an adjusted, modded player with a nice pickup... an LD-S2 might be out of my price range, but a CLD-1010 would be nice, and so would a gas tube model for Discovision disks...)

I'm surprised someone in Japan hasn't beat me to this, and if they have I'd definitely like to see it, it's almost certainly farther along than mine is!

(edited for a little more detail and clarity)


Last edited by happycube on 08 Sep 2015, 16:30, edited 1 time in total. _________________
Happycube Labs: Where the past is being re-made, today. [meep!]
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 01 May 2013, 05:42 
Hardcore fan
Hardcore fan
User avatar

Joined: 27 Oct 2011, 08:51
Posts: 1081
Location: Wichita, KS United States
Has thanked: 0 time
Been thanked: 0 time
We need some people with knowledge on making masters and maybe someone who could get some tech shop in Thialand to produce some writable discs that could be used on a VDR-V1000 but also playable on regular players or with a slight mod to the pickup. I have two working units that could be used to make laserdiscs but unfortunately they only play/write on special rewritable laserdiscs. If we could get more of theses discs made then put out a working mod to players maybe we could make some new titles. The discs are limited to 30 mins CAV, but maybe they could produce a dual sided disc from the original 1 sided VDM-V130.

Hey, it's a long shot but never hurts to try.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 01 May 2013, 07:04 
Advanced fan
Advanced fan
User avatar

Joined: 18 Apr 2012, 10:13
Posts: 784
Location: Australia
Has thanked: 3 times
Been thanked: 6 times
For the right price I would buy a custom made new movie released on Laserdisc. That would be awesome!
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 01 May 2013, 07:32 
Hardcore fan
Hardcore fan
User avatar

Joined: 18 Apr 2012, 18:02
Posts: 1468
Location: United States
Has thanked: 26 times
Been thanked: 19 times
What you're actually looking for is ODC's RLV system, which was basically an LD-R. A couple of weeks ago I found a couple at the local uni's surplus sale with promotional videos for the school, and one of them (from 1994?) actually played well. The other, a couple of years older, had about as much speckling as a bad Discovision disk.

Those were purely industrial systems, very expensive and disks cost $300/ea, apparently. I doubt there's a functional one left in the world, and probably difficult to impossible to get one online.

If the wavelength was similar enough, one could concievably burn LD signals onto a CD-R and hope to recreate CD video that way, iff the player could be fooled into thinking a CD-R was a CDV disk. With only about 5 minutes of video per disk, it would be incredibly impractical for most actual uses, however.

---

The stuff I'm playing with could potentially lead to a new Laserdisc *player* (or more likely a very slow ripper), assuming you found a good way to read the pits off the disk and construct the FM signal off it. I figure while there are still working decks, however, it's far easier to capture the FM signal off the player and decode it, and with better math than I actually know how to do and good ADC capture, quite possibly improve on quality, for very nice preservations.
_________________
Happycube Labs: Where the past is being re-made, today. [meep!]
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 01 May 2013, 09:19 
Hardcore fan
Hardcore fan
User avatar

Joined: 27 Oct 2011, 08:51
Posts: 1081
Location: Wichita, KS United States
Has thanked: 0 time
Been thanked: 0 time
I have a few RLV discs, but they're essentially worthless for recording. There was never a RLV recorder in the same sense as a Pioneer VDR-V1000 or some of the other set tops. There were several RLV outlets across the country but these were custom labs and the only model number I found was the Nimbus 610, which I believe was a very large industrial machine, like you said. They've probably completely dismantled those systems by now, unless someone kept one, but without discs I don't know why.

As for decoding the signal from the player, how much better do you think the output would be direct from the signal vs the player's output ports? I wouldn't think it would be much difference, if any. And where exactly are you interfacing with the player?
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 01 May 2013, 17:32 
Hardcore fan
Hardcore fan
User avatar

Joined: 18 Apr 2012, 18:02
Posts: 1468
Location: United States
Has thanked: 26 times
Been thanked: 19 times
I'm interfacing to an RF output test point.

And it depends on the quality of the RF signal (which would be improved greatly by an AC3-style mod), and the quality of the digital signal processing. If everything is done well, I think it could match if not exceed the best players. There's a lot you can do in a core i7 that 1990's designs couldn't, especially if you're not worried about pesky things like real-time decoding.
_________________
Happycube Labs: Where the past is being re-made, today. [meep!]
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 03 May 2013, 11:03 
True fan
True fan
User avatar

Joined: 11 May 2009, 11:12
Posts: 271
Location: Norway
Has thanked: 0 time
Been thanked: 0 time
happycube wrote:
I've been working on decoding Laserdisc video directly from the RF signal, bypassing most of the player electronics and hopefully (eventually) getting cleaner pictures.

I've been using an LD-V8000 with the RF test point hooked up to a video capture card using cxadc for raw 8-bit 28mhz capture (which I updated for Linux 3.5) to get the RF signal, and wrote some very slow, dirty, code to decode the laserdisc image, and another program to do time base correction and NTSC color decoding (which amazingly hasn't actually been done by anyone yet, as far as I can tell... I was rather surprised by this)

I've posted the rough source to github: https://github.com/happycube/ld-decode - I did a little bit more work on this right now, before I head off for my real job ;)

There's a .jpg there with color bars with correct colors, albeit fuzzy.

The next steps are to clean up the software decoding (I'm thinking locking onto the analog audio signals, subtracting them from the RF, and then working out some sort of proper PLL for the video rf), combining the NTSC decoding phase using more information from the LD decoding, actually properly adjusting my players, and an AC3 mod but taken from the core RF signal.

(and/or purchasing an adjusted, modded player with a nice pickup... an LD-S2 might be out of my price range, but a CLD-1010 would be nice, and so would a gas tube model for Discovision disks...)

I'm surprised someone in Japan hasn't beat me to this, and if they have I'd definitely like to see it, it's almost certainly farther along than mine is!

(edited for a little more detail and clarity)


SUPERB! project! i have thought about doing this my self for some time, i've been also coding on a fpga for temporal denoising, which would be the best bet for cleanup, only problem is its heavy cpu usage, and optimizing is hard to get it fast enough for 30 fps real time.
_________________
Pioneer KURO KRP-500M, Crystalio 2 VPS3300, Lumagen 2144, HLD-X9
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 03 May 2013, 13:54 
Serious fan
Serious fan
User avatar

Joined: 14 Oct 2012, 13:21
Posts: 199
Location: United States
Has thanked: 0 time
Been thanked: 0 time
Consider myself a supporter of this project. :thumbup: There definitely needs to be a better way to digitize LDs outside of capture cards.
_________________
Blu-ray Collection | LaserDisc Collection
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 03 May 2013, 17:42 
Jedi Initiate
Jedi Initiate
User avatar

Joined: 14 Jan 2010, 09:44
Posts: 2864
Location: United States
Has thanked: 20 times
Been thanked: 58 times
This is very interesting stuff. I've been thinking for the past year or so that in the future the best way to play an LD would be a combination of what you're working on and this: http://www.bbc.co.uk/news/technology-22294017
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 03 May 2013, 19:43 
Advanced fan
Advanced fan
User avatar

Joined: 02 Oct 2008, 04:35
Posts: 586
Location: Minnesota, United States
Has thanked: 0 time
Been thanked: 0 time
I saw that BBC story when it came out and thought it was really cool. Isn't that pretty much a modification of a laser turntable though? We already have a laser in our players :P

If you could use a camera to capture the whole sequence of pits and flats and convert that way, that would be really interesting. Images to audio video.
_________________
Laserdisc Player
CLD-3080 w/ AC-3
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 03 May 2013, 23:34 
Jedi Initiate
Jedi Initiate
User avatar

Joined: 14 Jan 2010, 09:44
Posts: 2864
Location: United States
Has thanked: 20 times
Been thanked: 58 times
Yeah, that's my point. Take a 3D map of the entire surface in one scan and then software extrapolate what the raw signal would be and software rasterizer the image.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 04 May 2013, 19:52 
True fan
True fan
User avatar

Joined: 21 Jun 2010, 20:03
Posts: 433
Location: TN, United States
Has thanked: 2 times
Been thanked: 1 time
Whoa, this could change everything for those of us who try to preserve films and certain cuts having to work from LD and the best players we have. Awesome idea.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 05 May 2013, 00:04 
Hardcore fan
Hardcore fan
User avatar

Joined: 18 Apr 2012, 18:02
Posts: 1468
Location: United States
Has thanked: 26 times
Been thanked: 19 times
I've been a bit busy this week, but I've made a bit of headway - the LD decoding code is now using slope detection. The quality isn't quite as good yet, but it's much faster. I'll post it after some more tweaking and/or the next evolution (it's slowly becoming a Kalman filter, I think)

I also just got a Rigol scope with 1M sample capture memory, I haven't had time to wire it up yet, but that'll give me better raw samples for the earlier phases (and if I do my own AC3 mod, let me *test* it)

I'm about to go on a big road trip, so I'd better have some good samples for it... afterwards, I'll get a player better setup for continuous capture.

krbahr et al: Is the D704's AC3 output full range or just audio RF? The service manual's sadly not online, and if it's full range I'll be interested in getting one in a couple of weeks.

- As for direct disk reading, I'm thinking a record turntable with an adapted USB webcam/microscope on a motorized system precise enough to track the disk. Assuming adequete sensitivity and magnification, a red LED (not laser) may be enough. At 30fps it might have to be *massively* slowed down even from 33rpm to get coherent information, but it would be a good presevation system.
_________________
Happycube Labs: Where the past is being re-made, today. [meep!]
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 09 May 2013, 08:20 
Serious fan
Serious fan
User avatar

Joined: 15 Jan 2013, 22:44
Posts: 135
Location: Canada
Has thanked: 0 time
Been thanked: 0 time
I think I'm in love.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 04 Jun 2013, 19:27 
True fan
True fan
User avatar

Joined: 08 Jan 2013, 18:13
Posts: 334
Location: United States
Has thanked: 0 time
Been thanked: 3 times
happycube: I sent you a PM.
_________________
Fill my eyes with that DiscoVision!
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 22 Jun 2013, 17:20 
Hardcore fan
Hardcore fan
User avatar

Joined: 18 Apr 2012, 18:02
Posts: 1468
Location: United States
Has thanked: 26 times
Been thanked: 19 times
Posted some updated code on github - still (pre?)alpha and I don't have an actual video encoding workflow yet. And the code itself is still a huge mess, as I keep trying things and not completely ripping them out ;) I'm just about to rip into the code a lot more, so I figured I'd post this iteration of it.

For me this has largely been about learning how to use digital filters - I'm going to try out some FIR filters next, along with breaking it up into per-line processing to support multithreading. And then hopefully rig up better RF capture...

An interesting side realization is that ghosting in Pioneer players is probably due to the analog filtering. The really nasty filter for LD might be the high-pass video filter at the beginning, since phase change varies with frequency. That could explain why the V8000 has it so bad, since they put a lot of effort into the filtering, ironically. This picture has 99 problems, but ghosting ain't one:

https://raw.github.com/happycube/ld-decode/master/rd-mono.jpg

And here's a turtle:

https://raw.github.com/happycube/ld-decode/master/turtle.jpg
_________________
Happycube Labs: Where the past is being re-made, today. [meep!]
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 18 Jul 2013, 00:39 
User avatar
Happycube,

Great project !

A few notes:
You definately need a copy of IEC60587 if you don't already have one, there is at least one gotcha that you are going to need to watch out for:
To cope with the large amount of analogue signal processing in the player the signal (mainly the low-pass filter) the video is group delayed at mastering time with the following spec (NTSC):
0.5 mhz 0ns
2 mhz -15ns +- 15ns
3 mhz -45ns +- 15ns
3.58 mhz -80ns +- 15ns
4 mhz -135ns +-30ns
4.2 mhz +-200ns +-30ns
Depending on exactly where you are picking up the RF signal, you may already have this applied or not.

You should be able to see this on any disc with a multi-burst recorded, and may explain why you are seeing some odd phase shifts in the chroma.
You could probably calibrate the group-delay filter using the multi-burst on a frame by frame basis...

Here are a couple of web pages you should definately read:
http://www.jim-easterbrook.me.uk/pal/
http://privatewww.essex.ac.uk/~tim/DIG_PAL/

Fourier composite decoding is probably the way to go to get better performance over existing 3D filters.

I have an idea for a way to implement an improved 'PAL Transform' style comb filter that would work for both NTSC and PAL, and give 3D comb filter performance with virtually zero artifacts but it is pretty computationally expensive (even compared to comb filtering using optical flow) and would be unlikely to work in realtime. I'll have a bash at implementing it in my mother tongue of Python.

I've been looking out for a streaming 0-100Mhz ADC (ideally USB3!) for this sort of project for years...

All the best.
  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 18 Jul 2013, 21:42 
Hardcore fan
Hardcore fan
User avatar

Joined: 23 Sep 2003, 18:14
Posts: 1391
Location: United States
Has thanked: 38 times
Been thanked: 17 times
Sounds like implementing a MUSE decoder would be much simpler!

That's been something of an idea of mine for some time.
_________________
MUSE decoder information and user guides
LD player connexion guide
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 19 Jul 2013, 18:39 
Hardcore fan
Hardcore fan
User avatar

Joined: 18 Apr 2012, 18:02
Posts: 1468
Location: United States
Has thanked: 26 times
Been thanked: 19 times
Thanks! I'm finding it very interesting, and I'm learning a lot about DSP along the way.

Digital low-pass FIR filters also have phase effects, so the pre-delay might actually be helping me. LD was surely designed for an analog low-pass filter, so I should look at using an IIR filter again, perhaps, and hoping there's enough signal quality something with a similar phase response to be useful.

The pictures do not reflect the latest checked-in code, since it's very much in flux. Right now I'm using a set of continuous fourier-ish transforms at different frequencies, running each through an 8-tap FIR, then performing phase comparison. It's a strange mix of DFT and phase-locked loop stuff. The result is then run through a 7-tap FIR filter. With test data it gets about ~55-56DB (if I'm computing it right, which is unlikely), but with real RF data is probably in the mid/upper 30's because I'm not doing any explicit noise cancelling.

I'm interested but not terribly good at electronics, so I don't have the RF signal amplified yet so the video capture card is putting in quite a bit of noise. I'm not sure which noise is from the card and what's from the disk. Hopefully I'll have an amplifier circuit working soon. The AC-3 RF out circuit is different, it raises the base from 0v to 4.5v and keeps amplitude the same, I need to go from .3-.4V P-P to 1.0 or so for the capture card.

A conversion circuit from AC-3 RF out to what the video capture card needs is possible, but I don't have any working players with an output anyway. I have ~2-3 LD-V8000's (one needs a new gear that I can probably print at the hackerspace eventually, and I'm torturing the one I've been working with ;) ) and an S104, with a board donor for it in storage.

In any case, the raw data and LD processing isn't great, and it's terribly slow, but it's good enough to move forward on the TBC and NTSC color processing so some test clips could be made before going back and improving them.

I made some headway on TBC, but my attempt at making motion video was a bit disappointing, so I'm going to refactor things a bit (for one thing, so I can get audio :) ) and then hopefully get video working relatively soon, even if it is noisy.

MUSE would be interesting to do, but even compared to LD it's terribly niche and a MUSE player and disks are far too expensive for me. At least with NTSC decoding, a really good comb filter might have applications outside of LD.

And I would love to have a 100mhz USB3 ADC board. I was hoping one of the SDR groups would have one by now, but no such luck.
_________________
Happycube Labs: Where the past is being re-made, today. [meep!]
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 21 Jul 2013, 15:30 
Jedi Candidate
Jedi Candidate
User avatar

Joined: 29 Jan 2006, 20:05
Posts: 2249
Location: United Kingdom
Has thanked: 0 time
Been thanked: 8 times
Maybe I'm missing something here but I don't see what all the fuss is about in this thread.

happycube is still having to use a laserdisc player to capture the RF signal output so what exactly is the benefit here?
The only difference is the captures he is doing are far lower quality than even the most basic digital capture card would achieve these days.
Best case it might be possible to improve the quality but you are still 100% reliant on the laserdisc player to extract the RF signal from.

I think there is merit in investigating technology which can make laserdisc players totally redundant by managing to extract
the signal via some other means such as scanning the discs etc but not via the route happycube is proposing which is like using oil
to extract new oil at twice the price.
Offline
 Profile  
 
Display posts from previous:  Sort by  
 Page 1 of 22 [ 428 posts ]  Go to page 1, 2, 3, 4, 5 ... 22  Next


Who is online

Users browsing this forum: No registered users and 7 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  

cron