LaserDisc Database
https://forum.lddb.com/

(WIP) Laserdisc software image decoder from raw signal
https://forum.lddb.com/viewtopic.php?f=32&t=2671
Page 26 of 31

Author:  titan91 [ 17 Sep 2020, 23:04 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

9954tony wrote:
titan91 wrote:
Chad, would it be possible to add a feature to ld-analyse to export the frames of a TBC file as a series of PNGs? There's already a single frame export feature, hopefully it wouldn't be hard to add this in as another option. ld-chroma-decoder does the job, but it outputs raw bitmap files that must be further muxed and processed by ffmpeg. This extra option would really help with the learning curve and speed of conversion I feel.


This would probably be "really" low priority from them, because, as you said, you can already do this with a longer tool chain command line. I wrote a perl script that basically does this, download that. or here is the direct code:
#!/usr/bin/perl
$staticdiv = 1500000;  #very crude aproximation of frame size
if (!$ARGV[0]){
        print 'Must provide file name, number of pics, pal/ntsc as input\n';
        exit 0;
}
chomp($ARGV[1]);
$filesize = `ls -ail $ARGV[0]|awk '{print \$6}'`;
$frames = int($filesize / $staticdiv); #rough count of number of frames
for ($x=5; $x < $frames; $x=$x + int($frames / $ARGV[1])) {
                print "FRAME:$x\n";
                if ($ARGV[2] =~ /pal/ ) {
                    $command = qq(ld-chroma-decoder -s $x -l 1 $ARGV[0] -|ffmpeg -f rawvideo -r 25 -pix_fmt rgb48 -s 928x576 -i - -an -vf "yadif,scale=768:576" -y $ARGV[0]$x.png);
                    $output=`$command`;
                }else{
                    $command = qq(ld-chroma-decoder -s $x -l 1 $ARGV[0] -|ffmpeg -f rawvideo -r 30000/1001 -pix_fmt rgb48 -s 760x488  -i - -an -vf "yadif" -y $ARGV[0]$x.png);
                    $output=`$command`;
                }
}


USAGE: getpics.pl input.tbc 6 ntsc
That would output 6 .png files from input.tbc spaced equally throughout the capture. ntsc is default, so you could technically leave that off.



Cool, I'll try the script out. Thanks for that. I also started looking around on eBay and found this which claims up to 30dB with a low enough band to amplify the HF signals needed. It's similar to the one you linked to in the endless sea of inexpensive options from China. I also have some spare 50 ohm SMA cable and adapters. I can probably tap a 12v rail inside the player to power it as it draws less than 100mA.

https://www.ebay.com/itm/RF-Amplifier-M ... 3737109040

Author:  9954tony [ 19 Sep 2020, 01:35 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Some days are better than others.

Image

Got a really good capture today with the TV card. This card won't go into 10fsc mode, i hard coded it in the cxadc.c and recompiled and reinstalled and rebooted, still 28msps. This is 1050 though. No amp.

I turned up the RF gain pot all the way (if you turn it down too far, the picture drops out, so that is how i found out which way was "UP" )

Author:  titan91 [ 19 Sep 2020, 21:19 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Wow, that's the best result I've seen since the first RF chipmunk samples on YouTube. So this is simply using a BNC connection and 50 ohm mini coax going directly to your card with the RF gain pot on your player adjusted to max? If so are you using any gain on the Conexant card, or is that set to 0 to minimize noise?

Author:  9954tony [ 19 Sep 2020, 21:44 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

titan91 wrote:
So this is simply using a BNC connection and 50 ohm mini coax going directly to your card with the RF gain pot on your player adjusted to max?

short answer: yes

longer answer: The 1050 has a singular "PIN" on the RF board for the RF test point, not a plastic connector like later models. I found some old piece of wire that had the female version of that connector on it and soldered that to the coax. I think the coax i used was just a piece i cut off an old PCI wifi antenna, smaller diameter than a #2 pencil, but not by much. That runs to a panel mount BNC connector i mounted, with the coax soldered to that. From there, just 3' or so BNC cable. Then i have the S-VIDEO to BNC Y splitter thing.

titan91 wrote:
If so are you using any gain on the Conexant card, or is that set to 0 to minimize noise?

I ran the "leveladj" and let it choose the level, i think it landed around 17.

There are some light squiggles in blue and red colors still, that picture looks so good, because it doesn't have any or much blue and red. That tells me the squiggles are some byproduct of my player, and the AMP amplified them.

As i am unable to get this particular card to go into 10fsc mode, these are 8fsc. Maybe try that too?

Author:  titan91 [ 19 Sep 2020, 22:54 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Ok. Do you mind uploading thst sample as well with the bees?

Author:  9954tony [ 19 Sep 2020, 23:33 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

titan91 wrote:
Ok. Do you mind uploading thst sample as well with the bees?


My previous post was only a partial ld-decode (e.g. i did a "-s 9700" or so) of a pretty long capture. This file is 11G: https://ready.cs.washington.edu/~tander ... scmikro.r8

Let me know when you downloaded all the samples, as this is a shared server and i don't want to take up too much space for an extended period of time.

There are some scenes with a blue sky and clouds, where you can see the squiggles lightly.

Author:  titan91 [ 20 Sep 2020, 01:21 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Thanks. I am downloading the first 2 samples from your PM but due to the speed from your university, I will need to grab the 11GB sample tomorow. So for anyone else that wants this sample, you can get it now.

Author:  titan91 [ 20 Sep 2020, 02:34 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

I decoded Tony's 28.6 8fsc sample and compared it with my 35.8 10fsc one. He is using an LNA from eBay. His spectrum is on top and mine is on bottom.

Image

The signal is clearly showing higher, but I'm seeing what appear to be images or upper and lower harmonics. This may be the reason for some of the squiggles he is seeing. My signal looks more defined, but the video carrier is just about lost in the noise floor. Listening to them both in Audacity at a 2KHz playback rate, mine has heavy "roar" in the background. I'm also hearing harmonic distortion in my samples in the form of a new frequency that comes and goes during the vertical blanking interval. Kind of sounds like a lower frequency resonance. This is likely a reflection in the cable and connectors I'm using due to poor impedance matching.

Image

Tony's noise floor is very low and his signal level is about 20db higher. Some of that noise may be reflected to the lower band in my case, but I think most of it is my particular HP branded card, or the Radio Shack distribution amplifier. Considering I get better SNR results from my VCR due to a higher signal level, I can probably overcome that. Once I get 50 ohm mini coax and a BNC connection installed, I will remove the Radio Shack amp from my signal path and try again. If my results still suck, I will make adjustments to the RF gain potentiometer in the player.

Author:  9954tony [ 20 Sep 2020, 04:42 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

I don't know if it is worthy of note, but my card is slightly different. After installing the cxadc driver, it didn't work. Investigation led me to discover that another driver was being installed, that wasn't part of the cxadc blacklist. The driver that Ubuntu was installing for my card, was:cx8802, which i had to add to the list of blacklisted drivers before i could get cxadc to work.

Author:  9954tony [ 21 Sep 2020, 16:11 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

A few years ago, when i was trying to track down a PQ quality problem with a player i bought on ebay, turning "DOWN" the RF gain on the RF board reduced the artifacts on the screen. So arbitrarily turning up the player RF gain, could introduce problems.

Turning up the player RF gain may allow capture with CXADC at a lower TV card gain (and therefore a lower S/N in ld-analyse), but it may also introduce artifacts. It would be wise to annotate the position of the setting before experimenting. Or for the less cavalier, don't mess with it at all.

The player i am experimenting with is of little value or importance, so i do what i do with full knowledge and understanding that it will need a full calibration before any serious captures are made.

Author:  titan91 [ 22 Sep 2020, 21:08 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

This doesn't seem like it has been done before, so I'll try a 50 ohm SMA connector with some mini coax. I'm not able to install a full BNC connector due to the drilling needed and I'm curious to see how much this improves things.

Author:  titan91 [ 05 Oct 2020, 01:33 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

I ordered a male SMA to male BNC adapter cable using 50 ohm mini coax, along with a 25cm female BNC to male s-video mini DIN pigtail adapter. This will connect to the RG-174 cable and female SMA connector I will install in the player. Like you, I got this from a cheap wifi antenna I no longer have a use for. So now I have an ideal cable and connector, plus some spare cable if needed.

Will compare the results when I have everything together.

Author:  9954tony [ 05 Oct 2020, 03:01 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

i don't know if i mentioned this before, but i have two cxadc cards. one of them performs significantly better than the other. so there is variation between the card designs. they both have the same cx chip.

Author:  titan91 [ 06 Oct 2020, 13:39 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Thanks for that info. What are the model numbers of the cards? Mine is an HP 5187-4378 Asus PVR-416 Blackbird.

Author:  titan91 [ 08 Oct 2020, 00:52 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Does anyone in the US have any leftover blank/unpopulated D3D boards? I will buy one from you if it's cheap.

Author:  9954tony [ 10 Oct 2020, 02:54 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

I've been playing around with modifying the cx288n boards to improve performance. I have achieved some pretty significant results i think. Second opinion wanted. this is currently suboptimal, because the low pass filter i am using is too low, and is cutting off the higher frequencies i want (but it is what i had on hand). The terminator is 75ohm instead of 50ohm, because again, it was what i had on hand. I believe i have found, and i have ordered, a better low pass filter here: https://www.minicircuits.com/pdfs/BLP-10.7+.pdf <-- that is the inline BNC one, but they also make one you would have to solder onto the board or make a casing for, that is less than half the cost of the BNC one.

This started by realizing there was a CVBS frequency low pass filter built into the boards, this would adversely affect the capture of signals over 6mhz. I removed them on 2 of my boards, and noticed a big improvement, but also a big difference still between the boards. Upon investigating and tracing the signal path, i discovered that board (A) has very little circuitry from the starting point to the CX chip. Board (B) had a "switch" IC that allowed the card to route different inputs to different locations around the board, one being the CX chip. I then started thinking about bypassing that chip, then i had the idea, to just bypass everything. There is now only a 100nf DC blocking cap between where i soldered my lead and the CX chip. I removed the components on the other side, so the signal wouldn't go anywhere else. The card is modified with a BNC connector.

Image

I used a BNC cable/t connector/terminator/low pass filter.

Image

I achieved 42db s/n ratio with this capture
Image

Author:  titan91 [ 12 Oct 2020, 23:36 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Fantastic stuff! Do you mind posting a link to your latest B&B sample? I'd like to analyze the spectrum. It may help others as well. If you're still using your university's file server I can mirror it to my MediaFire account for a more permanent link and you can delete your copy.

Edit: Here's a 1GB sample for anyone that wants it:

http://www.mediafire.com/file/xto19ya8s ... bw.r8/file

Author:  titan91 [ 14 Oct 2020, 00:28 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

Here's my RG174 cable from an old wifi antenna.
I added some short wire so I can reach the test header on my CLD-V720.

Image

Heat shrinked to prevent shorts.

Image

It attaches perfectly. I was going to use pin 5 for ground but the pins are too close together for my comfort level.

Image

I snaked the mini coax under a factory cable holder and used a zip tie to hold the SMA female end in place. I would prefer BNC, but I'm not about to attempt to drill in to the case and leave conductive metal shavings everywhere. Plus I would probably screw up the machine.

Image
Image
Image

And here's the new 5ft RG316 cable.

Image

I just need the pigtail adapter and will update with the results.

Author:  titan91 [ 16 Oct 2020, 18:37 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

So I just realized the spectrum plot in Audacity defaults to logarithmic scale. Using linear scale makes the spectrum much easier to see. It's a drop down menu.

Author:  titan91 [ 20 Oct 2020, 02:50 ]
Post subject:  Re: (WIP) Laserdisc software image decoder from raw signal

I also decided to get 2 cards very similar to what 9954tony is using. These were listed as being similar, but the photo of one of the listings was incorrect. As a result, they are both identical. This is great for comparing the stock performance of both cards. The spectrum and SNR values of the two are identical as a result (at least with VHS, I would expect the same for LD).

https://i.imgur.com/VpZRLYW.jpg

https://i.imgur.com/O6UgkUU.jpg

The plan now is to modify them exactly the same way with BNC and RG174 cable going to a DC removal capacitor and 50 ohm resistor to ground to match impedence. This will connect to one of the IC's inputs. The only difference between the two will be the low pass filters used.

Page 26 of 31 All times are UTC [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/