Star Wars Laserdisc Preservations. See 1st Post for Updates. — Page 21

RandomHajile said:

Hi there, first post but wanted to say that would it not be better if the capture from which ever LD player is made from the COMPOSITE video output...

and not the SVHS/S-Video output.

as in the video stored on laserdiscs is composite video.

and using the S-Video outputs are just the composite video being processed by the LD players late 80s/early 90s comb-filter tech...

so the best capture chain would be the composite video direct from the player (ideally modded with a true 75ohm BNC video connection) instead of capturing a processed S-Video output that will be processed again??

i have a 925 (and 909 and theta player) and as good as those S-Video outputs were, it all depended on if your TV had a weaker Composite video comb filter.

thanks tho for this thread as watching SW on laser ticks some boxes for old timers like myself :)  

There are many other individuals here who are much more qualified to talk about this than I but I'll try to answer you the best I can based upon what I've learned over the past few years.

When doing a traditional laserdisc capture which involves using a video output from the player plugged into a capture card it entirely depends on the type of laserdisc player you have.  If I was using an X0,X9, or R7G I would capture the s-video output because that probably is going to give better results than any capture card.

Anything less than these three players and I would use the composite output and plug it into the best capture card I could find and see what happens.

The other possibility is to use Happycube's software which basically uses a modified driver to turn some capture cards that use the Connexant cx23880 chip into a sort of software defined radio that can capture the entire rf signal directly from a test point and digitize it for later use.

He also has developed a software decoder which can take the digitized signal and decode everything entirely in the digital domain. It is currently a work in progress but already produces very good quality that rivals even the top players.  This has the most potential for making the best laserdisc preservations imo because you can use basically any laserdisc player and achieve high quality results.

Okay, I'm not disputing the fact that video is stored on laserdiscs as composite video. 

I'm aware of the basic signal flow inside of a typical laserdisc player. The fact remains that the x0,x9, and r7g will pretty much always produce the best real world results when capturing via the s-video input as long as you have the comb filter settings right when capturing.

I'm not sure that you understood that when I say that the rf signal is captured from a test point using Happycube's method that this means you are intercepting the signal after it has left the pickup but before it hits any analog filters much less the digital tbc/comb.

If you come up with some good results please feel to share them here with us.  I know there was a guy on another forum that had posted the exact same method you describe of tapping the digital tbc output on a player and running it into a sdi converter then to a pc.

He never posted any results though and that kind of soldering is typically beyond the average joe such as myself so that kind of mod will probably not be attempted by me unless it turns out to be the absolute best possible way of capturing a laserdisc lol.

On another note: Rotj-1478-85 is gout-synced. I just need to figure out how to crop the subtitles and include them in the black bars of the final encode.  I've been working a lot lately or this would have already been done.  Hopefully by next week it'll be on usenet.

If I'm not mistaken any of the newer LD players' composite output is still receiving the digital processing.  The players split the signal to Y and C and perform noise reduction, then recombine the signal to composite.  In fact most players just recombine the S-video signal to composite.  The signal does not come from the demodulator directly.  Only the X0 has the option of a clean signal path with its BNC connector, but this still has a TBC applied to it.  

RandomHajile said:

no doubt then a pro player from say sony/pioneer then would be a better/purer option then

and would have a BNC output :)

am sure a modern composite to sdi converter would have the requisite(improved) time based correction applied with a genlock input prob even more ideal.  

 I think it's pretty well established that the X0 provides the best image quality.  It's debatable whether it's BNC composite provides the better image than its comb filter through the S-video connection.  I know I don't have the cash to find out first hand.  

That's the benefit of happycube's software.  While a better laser would be preferable for reading the disc surface, it's possible that more common and cheaper players may offer similar (or better) quality when decoded this way.

In my opinion, that is the future of this hobby.  

RandomHajile said:

then you could process that digitally (frame pull down im guessing)

I'm truly sorry for your loss.

I'm glad you were able to solve it. I know I'd have still been stuck counting fields.

I'm all about strict conformance so I guess there will be some black frames introduced to make it all fit right. It will be the same situation with the Japanese P&S releases since they have seemingly replaced the binocular shots with 29.97 fps binocular overlays onto the 24fps film footage for whatever reason.

I will definitely enjoy being able to finally watch the time-compressed version without stuttering or interlacing though that's for sure.  Thanks again for your time and effort!  I'll try to get it all put together and on usenet asap!

Those three fields really do compose a single frame (GOUT 148551),
but the 3rd field is mis-aligned.
[Actually the first two are mis-aligned from the remainder of the shot.]

So we'll avoid using that field:

She said she said:

c148543=\
cSide2.trim(149323,149342).doubleWeave().pulldown(0,3)#(1,3) #148550
c148551=\
cSide2.trim(149343,149344 ).weave().assumeFPS("ntsc_film")#148551

I'm in the middle of the french thx release and have a question about audio. I'm assuming that for this release the audio needs to be slowed down to 24fps but retain the same pitch as 25fps.  What is the best/easiest way to achieve this?  Is it possible in avisynth?

Apparently there's an Avisynth filter called "TimeStretch" that can do it? No idea if it's any good.

Most PAL sources were not pitch corrected.  Are you sure you can't just slow down the audio?

"source.AssumeSampleRate(42294).ResampleAudio(44100)"

schorman13 said:

Most PAL sources were not pitch corrected.  Are you sure you can't just slow down the audio?

"source.AssumeSampleRate(42294).ResampleAudio(44100)"

 I'm not sure.  I just know that this release has been dubbed in french so I just assumed that they have recorded it at 25fps because of that.

To be honest, I have a hard time hearing the difference between the pitch adjusted version and the normal slowed down version but I know that other people have much more discerning ears than me so I figure I'd better try to get it right.

It is a 5 capture average and has been upscaled to 720p whereas the BD25 was a single capture and 480i. It also includes the japanese subtitles in the bottom black bar.

dave88 said:

As an aside, can the Technidisc even be identified just using the LD #'s

Yes, it even says Technidisc right next to the mint marks on the inner ring. I have four good copies.

dave88 said:

AntcuFaalb said:

dave88 said:

As an aside, can the Technidisc even be identified just using the LD #'s

Yes, it even says Technidisc right next to the mint marks on the inner ring. I have four good copies.

 But there are other copies with the same exact LD # and sleeve that are not "Technidisc" pressings right?

LD# and sleeve, yes, but the mint markings are different.

One interesting thing about the Technidisc SWE is that it has baked-in rainbowing. That is, it has rainbowing baked-in to the master that can't be removed via the phase inversion that comb filters rely on.

A good example of baked-in rainbowing is on the Star Trek The Next Generation DVDs. No comb filter can eliminate them. They're part of the signal.

With that being said, prior to changing to my new LD capture setup (Pioneer LD-V8000 to an ADV7842 evaluation board), I captured all of my Technidisc SWE copies with my existing setup (Runco LJR-II to Leitch DPS-575) since a more advanced comb filter wouldn't help anyway and the Runco's undefeatable noise reduction does a really good job on its baked-in crosstalk.

Our best bet to remove the rainbowing from the Technidisc SWE is to low-pass filter the chroma. There's not a whole lot of effective chroma resolution in it anyway, so we don't lose much.

It's also worthwhile to note that almost all Technidisc pressings (including many other movies on LD) suffer from crosstalk and, as many LDDbers describe it, "chroma noise" which is the baked-in rainbowing I'm seeing, I suspect.

Since there is already a couple of transfers of the SWE Technidisc floating around in the ether I'm currently more interested in the PAL releases atm because they offer better y/c separation and a bit more detail than the ntsc releases. Also, like Antcufaalb says the rainbowing is baked in.  I think Mysycamore also posted some screens where there was substantial crosstalk happening too. Maybe that was just his pressing/ld player though.

Granted, my capture card has some issues with scene changes that have similar luma values but other than that it works quite well on PAL material it seems. Unless you do a stop-frame inspection it's pretty hard to catch with the eye even knowing that it happens.

EDIT:  Luma low-passing also seems to have benefits from what I've seen.  It definitely reduces noise and dot crawl without losing detail until you go below 420 tvl.  Pal theoretically should be ok at 440 tvl but I notice detail loss at anything under 500 so I don't go under that. 

Where can I track down the new JSC?