|
It is currently 20 Apr 2024, 04:11
|
View unsolved topics | View unanswered posts
|
|
|
|
someperson
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 08 Oct 2016, 18:33 |
Knows how to post |
|
|
Joined: 09 Aug 2016, 03:09 Posts: 9 Location: United States Has thanked: 0 time Been thanked: 0 time
|
I have everything set up now (trying to use an ATI TV Wonder Pro card), but I'm running into this problem when I try to run ld-decoder.py: Traceback (most recent call last): File "./ld-decoder.py", line 18, in <module> import fftfilt as fftfilt ImportError: No module named fftfilt
What do I need to install to get this module? I'm running Kubuntu 16.04 LTS. Thanks!
|
|
|
|
|
happycube
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 09 Oct 2016, 05:44 |
Absolute fan |
|
|
Joined: 18 Apr 2012, 18:02 Posts: 1614 Location: United States Has thanked: 71 times Been thanked: 88 times
|
oops! I never included fftfilt.py in the repository. check it out again now and see if it works... updates: I haven't found that PVR card yet (although I haven't done much digging either) and I'm looking at doing a much faster comb filter using TensorFlow. I think I've worked out the needed steps, I just need to write the thing
_________________ Happycube Labs: Where the past is being re-made, today. [meep!]
|
|
|
|
|
anonyme-x22
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 09 Oct 2016, 17:51 |
Genuinely interested |
|
|
Joined: 27 Sep 2016, 17:02 Posts: 48 Location: France Has thanked: 0 time Been thanked: 0 time
|
Hi,
Great works. Don't you think that 8 bit is a way to low ? Also wouldn't it be easier to take the decoded signal and put it on vga out ? From there you can convert it in HDMI! You can have these signals directly from the board: Y,C,R,V,B
VGA needs: R,V,B, Hhz , Vhz. Vhz is 50hz, or 60hz ! Hhz ... in fact the Vhz and the Hhz are optional (haven't tested without) Anyway you can use a video sync separator with the Composite sync to get the Hhz, and the Vhz
Rgds,
|
|
|
|
|
happycube
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 14 Oct 2016, 18:38 |
Absolute fan |
|
|
Joined: 18 Apr 2012, 18:02 Posts: 1614 Location: United States Has thanked: 71 times Been thanked: 88 times
|
As an aside, the Pioneer Special Edition LazydiscDVD of Rambling Rose is a direct port of the LD D2 master (complete with dot crawl color bleed in the credits, and some video-shot extra features...) - which could be useful as a comb filter color comparison.
The DVD was replaced by a Full Frame (hopefully Open Matte, but I'm not bothering to find out) edition three years later, but the 1999 issue is still easy enough to get - I just got a sealed copy off eBay.
It's also a case in point re how much nicer LD packaging could be - the PSE LD came in a box with a little booklet. On the flip side, it also shows how much more convenient DVD could be, since the last few minutes of the movie are on a third side... and the disk survived shipping with much less prep than the LD would take! Gotta appreciate the finer points of everything, really.
Last edited by happycube on 15 Oct 2016, 02:11, edited 2 times in total.
_________________ Happycube Labs: Where the past is being re-made, today. [meep!]
|
|
|
|
|
confederate
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 14 Oct 2016, 20:50 |
Advanced fan |
|
|
Joined: 23 Jun 2013, 02:37 Posts: 724 Location: Germany Has thanked: 116 times Been thanked: 59 times
|
happycube wrote: As an aside, the Pioneer Special Edition LazydiskDVD of Rambling Rose is a direct port of the LD D2 master (complete with dot crawl in the credits, and some video-shot extra features...) - which could be useful as a comb filter color comparison.
The DVD was replaced by a Full Frame (hopefully Open Matte, but I'm not bothering to find out) edition three years later, but the 1999 issue is still easy enough to get - I just got a sealed copy off eBay.
It's also a case in point re how much nicer LD packaging could be - the PSE LD came in a box with a little booklet. On the flip side, it also shows how much more convenient DVD could be, since the last few minutes of the movie are on a third side... and the disk survived shipping with much less prep than the LD would take! Gotta appreciate the finer points of everything, really. That's very interesting. Are there any other similar DVD releases ? I really want to get that DVD myself !
|
|
|
|
|
happycube
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 15 Oct 2016, 02:08 |
Absolute fan |
|
|
Joined: 18 Apr 2012, 18:02 Posts: 1614 Location: United States Has thanked: 71 times Been thanked: 88 times
|
I'd look for non-anamorphic widescreen 1998-2001ish releases. For R1 at least, many Disney and Paramount (Star Trek VI 1st issue, Mission Impossible...) and at least some Fox (Speed!) - most of WB's early ones were anamorphic/FS DVD-10s tho. I'm going to keep an eye out for a D1-mastered Lazydisc without too much change in DNR etc... edit: I'm not seeing any composite telltales on Speed so far, and it's non-anamorphic. Score! --- more-on-topic-edit: Found another CX23883 card - ebay #282202228173, pchdtv atsc tuner card. I don't know how good the video input is yet, of course, but there are five more of them at least for ~$20/ea shipped. And there's another seller with 10 (but sadly low-profile brackets) for $50 shipped... --- edit2: Got a prototype of a 2D CF in TensorFlow working today: https://github.com/happycube/ld-decode/ ... ents.ipynb, sample image included. Nothing special quality-wise but works at about 1.5X real time w/o optimizations on my rig...
_________________ Happycube Labs: Where the past is being re-made, today. [meep!]
|
|
|
|
|
msgohan
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 22 Mar 2017, 06:26 |
Serious fan |
|
|
Joined: 15 Jan 2013, 22:44 Posts: 136 Location: Canada Has thanked: 5 times Been thanked: 1 time
|
Trying ld-decode itself for the first time. Sadly, the result I'm getting looks severely messed up. It appears to have bearding, among other issues. Two sample raws, if you can take a look. The first is with leveladj auto-set; I think it picked around 21. The second is with it manually set to 1. Both produce the same issue upon decoding, though. I'm just using my lazy alligator clip setup, and because the clips end up touching adjacent pins on that connector shown in the Google Doc, I had to attach them elsewhere on the CLD-S201 board. The two points are labelled RF and RF GND, though. Is this just not going to work?
|
|
|
|
|
nemesis
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 23 Jun 2017, 00:28 |
Shows curiousity |
|
|
Joined: 18 Jul 2016, 05:18 Posts: 28 Location: Australia Has thanked: 0 time Been thanked: 7 times
|
Hey, great work with all of this. I was wondering if I could I get some help/advice? I'm trying to rip the analog video stream off MegaLD's for the Pioneer Laseractive. I can capture the analog and digital audio tracks with my current setup, and I'm trying to find the best way to capture a high quality video rip that preserves all the extra data in the vblank region. After experimenting with quite a few different methods, and not being happy with any of them, I'm looking at the work you've done here as a way to capture the video data from the Laserdiscs. I've got a capture card that should be able to do the raw RF capture you've described, but so far I've struggled to get the Linux software environment setup correctly. This is no doubt simply because of my lack of experience with Linux, but I've hit a bit of a wall with it so far. As a first question, can someone direct me to a particular version of a particular distro that they know from experience will work correctly, if the steps in that google doc are followed? Can I just download the latest version of Ubuntu and use that? There's also another angle I wanted to ask about. I'm interested in taking a slightly different approach to what you've done here. Rather than ripping the raw RF signal from the player and demuxing in software, then decoding the video, I've found the Pioneer Laseractive has an 8-bit ADC (MB40568) that processes the composite video signal and spits out digital data. I've been able to capture this full digital signal by chaining a few Saleae Logic 16 clones together and continuously streaming the data out in sigrok. What I like about this approach is I've used the original player circuitry to demux the analog RF signal and separate the video stream, so I don't really have to worry about noise or anything else. Once the data leaves this chip, I've got a video signal which I know is as good as the one the player itself was working with. What do you think of this approach compared to the RF capture method you went with? Also, can I feed this composite video signal directly into something from your "ld-decode" repo to generate a video stream with the full 525 lines per NTSC frame? Any help is appreciated.
|
|
|
|
|
simoni
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 11 Sep 2017, 14:42 |
Honest fan |
|
|
Joined: 11 Sep 2017, 14:24 Posts: 88 Location: Sweden Has thanked: 2 times Been thanked: 36 times
|
I realise that I'm a little late to the party, but I'm interested in performing a software decode of some rare PAL laserdiscs from the BBC Domesday project in 1986. You can read more about the background on my website http://www.domesday86.com/In order to do this I want to have as good quality as possible; I read in the documentation and forums that 8-bit capture is so-so... Therefore (since I'm an electronics engineer) I have built my own USB 3.0 based ADC. It's still very much a prototype at the moment, but I intend to release the design and source-code as GPL once I'm done. My set is is a PAL SONY LDP-1500P laserdisc player which I have tapped the RF test point to a BNC connector. This is connected directly to the ADC which has a 300mV peak-to-peak sensitivity to match the RF output amplitude from the player. In turn this is run through a 10-bit ADC sampling at 29.538 MSPS (an odd number I know, but it's to do with the attached processor - it runs at 384Mhz and my ADC clock generation is therefore 384/13). This ADC is connected to a Cypress Superspeed explorer board which provides USB 3.0 connectivity with enough bandwidth for realtime DMA over USB and then capture to disk. I've now completed some initial testing of the USB 3.0 ADC and I have a 50 second capture from a PAL CAV laserdisc. I know that the github version of ld-decode is expecting 8-bit/27.5MSPS capture; so my first question is: what do I need to do to 'explain' to ld-decode about my input file (which is a raw sequence of 16-bit unsigned numbers). Being 10 bit, the 0V reference of the capture is 512 (rather than 128 for 8-bit) - so I'm guessing I need to shift the values around in the file before attempting any decoding? Absolutely any help and/or advice would be greatly appreciated!
|
|
|
|
|
happycube
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 12 Sep 2017, 06:41 |
Absolute fan |
|
|
Joined: 18 Apr 2012, 18:02 Posts: 1614 Location: United States Has thanked: 71 times Been thanked: 88 times
|
Awesome I think if you convert it in the early input stages before the RF processing that would work nicely. I can also take a look at it if you put up a 1-2 second sample. And how fast can that ADC go? I'd love to see the best possible RF capture that setup can do!
_________________ Happycube Labs: Where the past is being re-made, today. [meep!]
|
|
|
|
|
simoni
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 12 Sep 2017, 07:32 |
Honest fan |
|
|
Joined: 11 Sep 2017, 14:24 Posts: 88 Location: Sweden Has thanked: 2 times Been thanked: 36 times
|
The ADC is a TI ADS825 which can handle 40MSPS at 10-bit with a 1V peak-to-peak sensitivity. The issue with the speed is getting it across to the PC without losing any data. USB 3 will support assured transfer or assured bandwidth, but not both In my testing so far I tried to run at the full 40MSPS but got some drop-out over the 1 hour test cycle. Lowering the throughput to 30MSPS seems to provide flawless transfer. This is more a limitation of the Cypress FX3 than the ADC. I'm using a DMA transfer through the FX3 based on 16-bit ping-pong buffering to transport the data without needing the FX3's ARM processor involved (it's a feature called GPIF based on a simple state-machine design). The whole thing is very much a prototype. Yesterday was the first time I got actual data throughput (which is why I decided to post here and start asking about what to do with the data). Although some analysis of the RF treatment stage has shown some issues (the opamp driving the ADC is unintentionally acting as a high-pass filter and needs some redesign). Once I've got the ADC circuitry working correctly I will try to modify the firmware to work at exactly 30MSPS (to ease the math required on the software-side). Basically though, the design works; so, once I have (what I think is) valid data capture I'll post a sample of it (it throws around 3Gbyte data a minute though... so even short samples are quite big). As a side note, the same overall design could handle 16-bit ADC but when I did the initial cost analysis the chips start getting expensive; and I wanted to keep the overall cost reasonable. 10-bit is 4 times the resolution, so it should get the job done Thanks for the response and the offer of help; I'm happy to know that you are still interested. The Domesday project is an important piece of British computing history - and the discs are degrading fast. Getting high-quality video from them is important; but having a backup of the physical discs is even more so as they are LV-ROM format and also contain digital data in the audio bands.
|
|
|
|
|
happycube
|
Post subject: Re: (WIP) Laserdisc software image decoder from raw signal Posted: 13 Sep 2017, 05:44 |
Absolute fan |
|
|
Joined: 18 Apr 2012, 18:02 Posts: 1614 Location: United States Has thanked: 71 times Been thanked: 88 times
|
I haven't done anything on ld-decode lately to be honest, but a new ADC used to capture the Domesday disks is very interesting Don't worry about getting exactly 30MSPS - disk speed is not 100% regular, and all existing PAL RF captures aren't even nominally a pure multiple of either the pilot signal or PAL color carrier. So go for the highest you can without any dropouts. Also note my code does not have any EFM decoding capability, so you'll want to tap the decoded digital data. The RF encoding is rather... odd and since I figured anyone who really wanted to preserve the digital tracks could with existing methods, I never put in the time to crack it. Ironically though, it would probably be more interesting to a lot of people since it would provide provably correct CD rips.
_________________ Happycube Labs: Where the past is being re-made, today. [meep!]
|
|
|
|
|
|
|
|
|
|
|
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
|
|