Laserdisc BEST source (PAL vs NTSC)

[darkbluesky]

Hi,

First, excuse me as I am not really sure if this is the correct part of the forum to post this (please moderators, move if it should be in other section).

I am going to make some LD captures for preservation of some classic Disney films (the DVD/BD editions are too much altered from the original), I would like to go for the maximum detail/quality as my first goal is preservation. So, I want to end with a capture with the maximum quality I could, and in progressive.

As discussed in other topic here, I'll capture in 720x480 NTSC (or 720x576 PAL).

If I understand correctly PAL version would allow me to have more vertical resolution than NTSC, but aside of that, and in general, are NTSC Laserdisc editions more advisable for capturing than PAL ones? or vice-versa? Normally which one you would search for if you were going to capture it? Resolution, color information/richness (is PAL worse?) are both important.

The NTSC versions I have are the USA CAV/Collector ones, and the PAL ones, are 'normal' (CAV/CLV?) in spanish. Some of the movies are Snow White, Pinocchio, Cinderella, Bambi, Sleeping Beauty, etc, which version -NTSC or PAL- would you recommend?

The player would be Pioneer CLD-D925, plugged directly to the capture card Asus MyCinema ES2-750 (ATI Theater 750) via composite, Lagarith YUV2 4:2:2 (or RGB24 if it is worth).

Thank you

[KarMa]

The PAL versions of these animated films will most likely be sped up film. So these films will play at 25fps instead of the intended 24fps. You can easily capture at 25fps, and then change the video playback to 24fps. But when they sped up the film, they also had to speed up the audio giving it a shorter duration and higher pitch. There are ways to correct the pitch and play time back for a 24fps playback but will probably have a bit of a loss, and probably won't be as good as the NTSC audio. You won't really notice the higher PAL pitch unless you can switch between two versions quickly and compare them. If you are ok with the slightly higher pitch than go with the PAL.

Your NTSC LD animated films are probably telecined, meaning that the 23.976fps progressive frames can be restored after capture. And the audio should require no special treatment. But will have less vertical resolution than the PAL version, as you already know.

[darkbluesky]

For the original/english soundtrack I guess I could take the one from the NTSC versions, which afaik (correct me if I am wrong) does not need any time/pitch modification, as the NTSC versions are not speeded up.

But for PAL non-english soundtracks I would need either live with it as you say, or to speed it down, which makes me wonder: can I be 'sure' that they speeded up the audio and not re-processed it to keep the correct pitch at 25fps? I ignore which is the usual practice. If they only speeded it up I would not need to correct pitch; I would only need to speed it down, and pitch and time would be correct (so, no loss).

Regarding image quality, I don't know if PAL has inferior color information vs NTSC, I readed something about it but was not too clear to me. On the other hand, I guess that it depends on the master, process used (I read the NTSC were mastered usually better than PAL, but...¿?). I wonder how could I evaluate IQ, aside of capturing and looking at it, of course.

[jagabo]

You can't be certain whether the PAL or NTSC version is going to be better until you've looked at both. Often analog PAL video is created from analog NTSC video (and vice versa) using an analog standards converter. In which case the resolution advantage of PAL doesn't come into play. And those converters usually use field blending so you'll want to remove the blended fields too.

Hopefully this doesn't apply to those A-list movies and the PAL versions will be the better source.

[jagabo]

Keep an eye out for automatic gain pumping with ATI capture devices. You'll see that problem at scene changes where one scene is dark and the other brighter. But sometimes the brightness seems to change for no obvious reason. That makes them pretty useless in my experience.

[Skiller]

Yep, that's the reason I sold my Asus MyCinema ES2-750 (PCI-E). The brightness pumping at scenes changes just ruined the captures for me. I mean it's not that obvious most of the time but once you know it's there you cannot overlook it.

[darkbluesky]

Hi, I am afraid of commenting on this (touch wood!) but I have never seen the AGC problem with my 750. I have tested this afternoon another time during a long fragment of Cinderella (seeing LD through virtualdub), did not see even the smallest hint of AGC in scene change, contrast change (dark/light), etc.

I tested this also when, some years ago, I read for the first time about this, but never saw the problem...even looking for it. On the other hand, I would like to know which alternative we have for SD capturing, with good 3D comb filter, 10 bits or so and overall at least as good as the ATI, for uncompressed capture...

[jagabo]

You don't need 10 bit caps for laserdisc. It has only 5 or 6 bits worth of signal to noise ratio to start with. Capturing 10 bits instead of the usual 8 bits will just give you more precise noise -- which you will be filtering away later anyway.

[usually_quiet]

On the other hand, I would like to know which alternative we have for SD capturing, with good 3D comb filter, 10 bits or so and overall at least as good as the ATI, for uncompressed capture...

For PAL, the Hauppauge WinTV HVR 5525 is an internal TV tuner/analog capture device for the PC with a 3D comb filter for SD analog capture. It uses software to encode analog video so it is possible to do lossless capture. As a side benefit, it also has a versatile digital tuner capable of receiving several types of DVB signals. For ATSC + NTSC, the Hauppauge WinTV HVR 1250/1265 is the closest equivalent. Both probably output video using 8 bits per pixel. The decoder chip is 10-bit internally.

WinTV-HVR-1265: http://www.hauppauge.com/site/products/data_hvr1265.html
WinTV-HVR-5525 http://www.hauppauge.co.uk/site/products/data_hvr5525.html

[Edit] Data sheet for the analog decoder
http://205.175.208.33/servlets/DownloadServlet/PBR-200974-004.pdf?docid=975&revid=4

[lingyi]

Since you're seeking the best possible quality, consider getting another (or several if possible) players. Ask around, you may be able to borrow rather than purchase.

Like analog VCRs, picture quality varies by machine and discs that don't play well in one machine, will work in another. The exhatled Japanese release only HLD-X9 is reported to be able to play back discs with bad chroma noise flawlessly.

Unfortunately, most of the best LD players (including the exhalted HLD-X9) are NTSC only.

Good luck with your transfers!

[darkbluesky]

This wintv card 5525 sounds nice (thanks for datasheet link!), a pity that it only can capture PAL and not NTSC (a second card will be needed then, and the 1265/1250 is not available in europe), but anyway, thanks for the advice. It's much appreciated! For the moment I'll try to use the 750, for the reasons explained above. If I have the chance though, it would be interesting to get the wintv and do some tests.

I have two more Pioneer LDs decks (CLD-D515 & CLD-900S), more consumer grade, anyway the top japanese players are completely out of my reach right now, so if I would need some of them I should look for someone who'd lend it to me, eventually. I think that for PAL there is nothing better than a D925 or 2950, iirc.

On the other hand, about my comment on audio pitch if I use PAL (see post #3), what do you think?

[usually_quiet]

This wintv card 5525 sounds nice (thanks for datasheet link!), a pity that it only can capture PAL and not NTSC (a second card will be needed then, and the 1265/1250 is not available in europe), but anyway, thanks for the advice. It's much appreciated! For the moment I'll try to use the 750, for the reasons explained above. If I have the chance though, it would be interesting to get the wintv and do some tests.

The WinTV HVR-5525's analog tuner only supports PAL, but there is a chance the WinTV HVR-5525 can capture NTSC from its S-Video or composite connection, although I don't know that for certain. The analog decoder chip supports both NTSC and PAL.

[johnmeyer]

I think you are "over-thinking" this a bit. While better than Beta/VHS/8mm, laserdisc is, at best, equivalent to S-VHS. For reasons posted many times, you will not get any improvement from using the S-video output from your laserdisc player (it won't hurt to use it). Make sure you are getting the digital audio channel, not the analog channel. That can make a huge difference. I don't think it is worth getting a player that handles AC-3 audio, although others may disagree.

If you are concerned with quality, most quality improvements with laserdisc are those you do in post production. Unlike transferring from tape, I don't think you will get as much of an improvement from using a TBC, and I've certainly never seen any flagging or obvious timing problems on vertical objects, something that you almost always see in VHS captures done without a TBC.

As to whether the PAL or NTSC version of a particular movie will be better, I am pretty sure that you cannot make any blanket statements about that because it will depend mostly (as others have already said) on the chain of events that led to getting the video onto that particular disc. Since laserdiscs did not permit native 24 fps, the NTSC version of the film (or animation) will obviously be telecined, and you will need to remove that entirely before encoding the movie. So, you'll have to use TFM/TDecimate (or some equivalent). For animation, you may need to use some non-standard parameters for both of these functions.

For PAL, if you are lucky, it is merely sped-up 24 fps. You can simply change the playback speed in the header (if you capture to AVI there is a utility [AVIFrate] that will do this), and then adjust the audio by 24/25, using a competent sound editing app. When you do that, you'll need to decide whether you need to change the pitch. I think most transfers of that era probably just sped up the audio and lived with the slight increase in pitch. If so, you'll want to lower the pitch when you slow down the audio (that's what I'd do, if I couldn't find some clue as to the original pitch).

Since it is animation, even with PAL you may need to do some inverse telecine. However, I am not an animation expert, so others will have to help you with that. The only time I ran into that was when I bought a rare "Songs of the South" laserdisc from Japan (the film is considered controversial so Disney never released it on DVD, and did not even release the laserdisc version in the USA). It mixes a little animation into the live action. I just used the IVTC that I was doing for the live action, and the animation portion turned out just fine.

[edit]I just read a post from a few days ago which provides more detail about some of the issues I mentioned above:

Best capture for laserdisc preservation

[edit2]Oh wait, that thread is you again, posting about the same subject ...

[darkbluesky]

Hi, thanks johnmeyer for the detailed explanation. Yes, I think I'll do as explained. BTW, I tried a PAL Cinderella, and the capture was already in progressive, no need for IVTC, but other prints could be different. Anyway, I'll check and see what is needed for each one. About the audio, yes, I think I'll adjust by 24/25 and the pitch will become lower automattically (if I do not do anything to keep it unchanged).

There are a lot of useful details in your post for me, thank you, I'll try to take them into account. Sorry for the other post, I did two of them to try to keep subjects separate, one for capture resolution and other for the best source, although I see that both subjects are getting closer in some points and even mix subjects somewhat. Sorry.

Edit:what you mention about post production, is what I started to think on now, but I have still to learn all about which filters and steps should I done to improve the laserdisc capture, and what can I expect.

[jagabo]

you will not get any improvement from using the S-video output from your laserdisc player (it won't hurt to use it)

Maybe you know his particular players... But most capture devices have better comb filters than those found on most laserdisc players. If that's the case here composite captures will have much less dot crawl artifacting.

[johnmeyer]

you will not get any improvement from using the S-video output from your laserdisc player (it won't hurt to use it)

Maybe you know his particular players... But most capture devices have better comb filters than those found on most laserdisc players. If that's the case here composite captures will have much less dot crawl artifacting.

Yes. I think you did a great job describing this almost exactly a decade ago:

Is COMPOSITE just as good as S-VIDEO when recording from LASERDISC?

[themaster1]

Panup is best (ntsc+ pal) , see here -> http://originaltrilogy.com/topic/PaNup-or-how-to-upscale-PAL-NTSC-capture-and-live-qui...happy/id/15928

[johnmeyer]

Panup: fascinating concept. Registration might be tough, and there also might be some slight aspect ratio difference that could be difficult to deal with. For film, gate weave would swamp the registration.

Still, I love the thinking behind this.

[darkbluesky]

Yes, I have read all the pages linked with panup, although I realize it would, likely, very hard to do, I would like to give it a shot, just for testing (after all, I have not seen any test on cartoons), but I have not found any guide as how to do it in, for example, virtualdub or avisynth, only a short comment about avisynth which I can't understand too well.

[spoRv]

Happy New Year to everyone!

I'm the "inventor" of the PaNup technique; the discussion continue here: https://forum.fanres.com/thread-25.html

Everyone is invited to sign up at our fan restoration community!

[jmac698]

Just want to mention there's another way to do dot crawl removal which is perfect. With the telecine process, sometimes a field is repeated, however that field has the colour 180 out of phase, so if you just average the fields, you get perfect dot crawl removal. The only problem is, you don't get it for all the film frames. However, it can be a basis to interpolate or guide some other methods. What would be great is if there were two sources with different phases (or maybe one had a video edit somewhere changing the phase) so you could have enough to completely recreate the colours.

Likewise, there may be some coincidental method from the pal version, especially as someone said they were created from NTSC masters. I'd have to think and write out the formulas involved in that, but with some subtraction of the two sources you could probably get more perfect lines of colour.

Could someone summarize this pan technique so I don't have to read the whole thread? I can understand perfectly at a high level with a short description.

[jmac698]

ps nevermind. Yes I thought of that idea already Glad someone did it. And now I just gave you the colour portion of the same idea plus dot crawl removal.

Here's some other ideas I've done also:
-combine full frame with letterbox versions. I discovered something interesting, the frame is moving around constantly, so it has to be dynamic.
-combine several copies of the same source to reduce noise
-combine several airings of digital tv to merge the least compressed frames. For example with 24fps film broadcast in 720p at 60fps there's several copies of the same frame, but in mpeg2 I found that the first b from after an i frame of the same film frame had the best quality picture. Even simple averaging of multiple digital recordings looks a lot better. Same with analog recordings direct from tuner.

[clashradio]

Your NTSC LD animated films are probably telecined, meaning that the 23.976fps progressive frames can be restored after capture.

So even though the captured AVI has a frame rate of 29.997, you can render it @24? that will improve quality?

Also want to confirm using the rca Composite output is actually better than using the S-video output; because the video on the disc is in Composite format?

[jagabo]

So even though the captured AVI has a frame rate of 29.997, you can render it @24? that will improve quality?

Yes. Film is usually converted to video by duplicating fields turning 23.976 fps frames into 59.94 fields per second (packed as 29.97 frames per second). You can restore the original film frames with an "inverse telecine". Doing so will get you better compression in the end. And your playback device or TV won't have to deinterlace.

Also want to confirm using the rca Composite output is actually better than using the S-video output; because the video on the disc is in Composite format?

Yes the signal on the disc is composite. The composite to s-video converters found on most LD players are very poor. Most capture devices will have better converters. But you should try it and verify for yourself. Try a disc with a lot of saturated colors and sharp edges.

[spoRv]

The inverted-phase technique works, in theory... take a look at this thread (second post):
http://originaltrilogy.com/topic/Comb-Filter-Testing/id/15421/page/4

PaNup in few words... well, read this article, it will take no more than few minutes:
http://blog.sporv.com/panup-or-how-to-upscale-pal-ntsc-capture-and-live-quite-happy

Just a PaNup example: http://screenshotcomparison.com/comparison/63368
(OK, final result is denoised and have a grain plate applied, but you get the point)

[darkbluesky]

It is VERY interesting the inverted-phase trick, but it seems that there is no way to do that for all the frames, isn't? Or maybe someone has already found how to make a second capture with inverted phase?

(BTW, why this trick works?)

[spoRv]

I tried to make it work, but it never worked actually... don't know if it's something to do with the italian electric system...

Well, if someone will find a way to invert the phase, it will work for the whole capture.

[jagabo]

The phase inversion is part of how all the 3d dot crawl removers work. The problem is, blindly blending two fields/frames together creates double exposures whenever there is motion. The dot crawl filters first detect what parts of the pictures are static. They use 3d techniques on those parts. Then they have to use 2d techniques on the rest.

PAL video uses a four frame pattern. So you need to blend 4 frames if you want to use phase inversion.

LD will always output the same phase at the same frame because the composite signal, chroma subcarrier and all, is what's recorded on the disc. VHS might randomly pick a phase when it starts since the luma and chroma are recorded separately and are multiplexed together before sending them out.

[johnmeyer]

combine several copies of the same source to reduce noise

That is one of the earliest NR techniques developed. There is some amazing software, developed over at doom9.org, that helps you "line up" the multiple captures. You need this because one of the biggest problems in any of the techniques being discussed here is getting each of the sources to precisely, to the frame, line up with each other. This is complicated by multiple real-world issues that are not part of the pristine theoretical discussion. This includes, but is not limited to: different tape speeds for each capture (even from the same tape on the same machine); top-field & bottom field issues (there is actually no concept of "top field first" or "bottom field first" in analog video and instead the fields just come one after the other and there is never any "frame"); frame drops during capture; and more. So, even when you do multiple captures from the exact same tape on the exact same machine, those captures may not line up perfectly over the entire length of the program.

Having to sometimes slip the capture by one field is the problem I found to be the most difficult when I actually did this a few times, a decade ago. When it works, this multiple-capture noise reduction technique has the fewest side-effects of any NR I've done, but it takes a LONG time, both because of the multiple captures, and also because of all the work getting each capture to perfectly align on the timeline.

Also, it only works for noise generated during playback. Any noise that is on the original source material (low-light noise, for instance) won't be improved at all. More modern digital noise reduction techniques work just as well for both noise generated during playback, and noise generated during recording. For all these reasons, I would never consider using this technique except for a really high-paying job, or for something of great historical importance, neither of which comes my way very often.

[spoRv]

Once I used four different versions of Star Wars on laserdisc for one of my projects; it was pretty easy to align them, both temporally and spatially, and result was rewarding. Probably using tapes will be not that easy, though!