Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
24th July 2016, 22:37 | #1 | Link |
Registered User
Join Date: Sep 2014
Posts: 32
|
Help cleaning up laserdisc transfer?
I've got a few quick questions related to a laserdisc transfer I have. What I want to do is inverse telecine it, upscale it to 1080p and include it on a Blu-ray I am authoring. It will be a special feature, not the main feature on the disc.
I didn't do the capturing myself. However, I know the person who did capture it for me used a VERY high end laserdisc player. A Pioneer HLD-X9 to be exact. I know he used S-Video cables to a DVD Recorder. He sent me a DVD9. Anyway, I am curious what AVISynth filters you'd recommend to get the best quality upscaled version of a laserdisc transfer? Here is the script I am using at the moment: Mpeg2Source("C:\DM2_MeiAh_Gold_merged.d2v", cpu=6) Crop(6,40,-6,-40) AssumeTFF() TFM(mode=5, pp=0) Tdecimate(mode=0) SuperRes(4, .8, 0, """nnedi3_rpow2(2, nns=4, cshift="Spline36Resize")""") SuperRes(4, .8, 0, """nnedi3_rpow2(2, nns=4, cshift="Spline36Resize", fwidth = 1920, fheight = 1080)""") I've been experimenting with DAA() and NAA() commands to get rid of any remaining aliasing. I'd just like to know if there are any additional filters that you'd recommend for laserdisc transfers? Thanks! |
24th July 2016, 23:47 | #2 | Link |
Registered User
Join Date: Feb 2002
Location: California
Posts: 2,691
|
First of all, laserdisc is simply analog NTSC video, so whatever techniques you have used for OTA SD captures and VHS/Beta/8mm analog video will also apply to laserdisc.
I still do laserdisc transfers, and am returning one today to a client. Here are the issues I have had to deal with in the past: 1. Sparkle. Laserdiscs often had white "sparkles," similar the white dropouts you get with VHS tape, but more circular. Noise filters or dust/spot removers work best. 2. Levels. Pay particular attention to your histogram or scopes and make sure the levels are correct. Laserdiscs always seem to be a little low on contrast, at least with my transfer technique. 3. Audio. You said that the transfer was done with high-end equipment. The S-Video output actually may be worse than composite. I remember all sorts of posts about this (in Compuserve forums) twenty years ago. I never totally bought into the arguments, but most of what people said was that, unlike VHS where there is a huge reduction in dot crawl and other composite artifacts, such was not the case with laserdisc. However, I'm sure your transfer is fine, and you shouldn't worry about it. The only real advantage to "high end" laserdisc players is that the later generation were able to extract the AC-3 audio, if the laserdisc had it. Most of the early ones did not, and even late in the laserdisc life cycle, many still did not. However, if yours did, hopefully you will have access to the better audio. Whether they were able to tap into the AC-3 output or not, they definitely should have used the digital rather than the analog channel. The differences are not as great as those between VHS linear and Hi-Fi audio, but they are still pretty substantial. 4. Sharpening. I have never felt that sharpening for VHS/Beta/8mm produce acceptable results, but if you have a really good laserdisc transfer, you might want to experiment with a little sharpening. 5. Noise. The noise on a laserdisc is usually much less than what you get from VHS, so you shouldn't need to do too much denoising. I usually use a scaled back version of my MDegrain denoising, although I think people here may be able to suggest a better option. 6. IVTC. The last two laserdiscs I did were both old sports films, so they needed IVTC. The client is really, really sharp, and he caught some pretty significant errors in the IVTC that I had missed. The problems were that I used the default TFM/TDecimate settings. What I found is that the matching was failing quite a bit, and I then looked back at my engineering notebook and discovered that the cthresh setting can make a HUGE difference in the outcome. In my case, setting cthresh=255 fixed all problems. I suggest you look into this and play around with the setting and see if you can spot places where the IVTC fails. What you will see when it fails is what looks like a dropped frame, i.e., the motion will jerk forward too far from one frame to the next, but only once in a great while. I've got to leave now, but if you have specific questions, post again. Also, if you have some video to post, I'll take a look at it. Others will have to chime in with suggestions about up-resn'g, although there are literally dozens of threads about that very subject, so you'll probably get more complete answers by just doing some searching. Last edited by johnmeyer; 26th July 2016 at 17:08. Reason: typo |
26th July 2016, 07:30 | #3 | Link | |
Registered User
Join Date: Sep 2014
Posts: 32
|
Quote:
From what I have read, when doing laserdisc transfers, you should be using composite and not S-Video. However, the exception to this is a small number of super high-end players where using S-Video is better. I'm not exactly sure of the reasons for this, but I'm thinking that the HLD-X9 is one of those select few players where S-Video is better. It doesn't really matter anyway because it's the only source I have at the moment. I've got the audio from another source and it is a very high quality capture, so no worries there. I'll see if I can post a short clip of my video sometime in the next day or so. Thanks! |
|
26th July 2016, 07:38 | #4 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
that's absolutely bullshit, you want "dot crawl" and "rainbow"? go composite, and how about just "rainbow", go S-Video, both are toxic and should never be used YPbPr is the master race here. |
|
26th July 2016, 08:09 | #5 | Link |
Registered User
Join Date: Oct 2007
Posts: 135
|
Most people refer to YPbPr as Component, which is confusingly similar sounding to composite. I really hope that's all this was.
Incidentally, you can use good old composite cables for component hookups, assuming it's a full set with audio. Just remember which color went where, since you obviously aren't going to match a white or yellow plug into a blue or green socket. Analog cable is analog cable, if you're in a bind. |
26th July 2016, 08:41 | #6 | Link | |
Registered User
Join Date: Oct 2002
Location: France
Posts: 2,308
|
Quote:
I've often do the same thing (transform 480i@29.97 on 23.976p video and put it on Blu-Ray), but just upscale to 720p. If i could avoid the upscale i would, unfortunately the Blu-Ray standard accept only 480i@29.97 or 576i@25 for SD resolution. The lowest resolution at 23.976 the Blu-Ray standard accept is 720. If you want to do 23.976p video, my advice would be to upscale only up to 720p. From my point of view it's useless to upscale to 1080p, save your banditwh to put more thing on the Blu-Ray, or increase the quality for the same final file size. Personnaly, i think you'll gain nothing upscaling up to 1080p. It's just a little... "sad" that he has to put it on DVD, because you have MPEG2 compression. The few times i've done a laserdisc transfert, i saved the capture on a lossless file, and work with this. Last edited by jpsdr; 26th July 2016 at 08:44. |
|
26th July 2016, 08:43 | #7 | Link | |
Formerly davidh*****
Join Date: Jan 2004
Posts: 2,493
|
Quote:
So dot crawl and rainbow are inevitable, it's just a question of what you use to separate the video without making them much worse. |
|
26th July 2016, 08:49 | #8 | Link |
Registered User
Join Date: Oct 2002
Location: France
Posts: 2,308
|
It's a very long time ago i've read spec on laserdisc video, but i thought remember (but maybe my memory is wrong), that luma and chroma where, like on VHS, recorded separately. I remember that video information is stored in frequence modulation, that's for sure.
But, i'm sure that on my Laserdisc player (the only Pioneer high end with S-Video output avaible in europe), the S-Video output was better than the Composite output. Edit : After a little search, video was indeed stored in composite format, not Y and C separately, so thing are exactly how davidhorman said, my bad. Last edited by jpsdr; 26th July 2016 at 09:16. |
26th July 2016, 17:16 | #9 | Link | |
Registered User
Join Date: Feb 2002
Location: California
Posts: 2,691
|
Quote:
Is COMPOSITE just as good as S-VIDEO when recording from LASERDISC? Years ago, when I first start watching my laserdiscs, shortly after I also got my first S-VHS VCR, I did lots of experiments using both S-Video and composite connections. The differences out of my VCR were obvious, but there were zero detectable differences between S-Video and composite connections when watching my laserdiscs. So, from actual tests with my own eyes and my own equipment, I think the information is, in fact, correct. |
|
26th July 2016, 18:06 | #10 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
and hell, why can't people just take component video and be happy with it. cheap shit like composite and s-video, they need to be destroyed. to the guy who invented composite video, Fuck You! (and the guy who invented s-video, the guy invented interlacing, the guy invented chroma subsampling, the guy invented limited range, the guy thought 8bpc was enough precision) Last edited by feisty2; 26th July 2016 at 18:27. |
|
26th July 2016, 18:28 | #11 | Link |
Registered User
Join Date: Oct 2002
Location: France
Posts: 2,308
|
Don't forget that TV exist since a long time, and if some thing are like they are, it's because there was a very very good technical reason, and at the time, there was no other possibility, because of technical limitation. You don't know analog as you said, i know of analog, history of TV, why things are as they are. Don't forget that there is a thing callled "evolution", technic was not always as they are today.
|
26th July 2016, 18:39 | #12 | Link | |||
Registered User
Join Date: Feb 2002
Location: California
Posts: 2,691
|
Quote:
Quote:
Quote:
What's even funnier is that you seem to be saying that all this digital stuff is wonderful. The problem is that it involves its own tradeoffs. Today I am going to try to figure out how to save some wonderful drone video taken in Iceland's national park at 2:00 a.m. in the fog. The problem is that the video, while spectacular, is almost unwatchable because of how your digital video handles fog. The problem is that digital video has its own pesky engineering tradeoffs, and they can be as bad as anything in the analog world. To be able to record digital HD or 4K video in a camera that can be held by a human being (or carried aloft by a drone), it has to be compressed. And, because of the physics and math involved, it can no longer be compressed one frame at a time, but instead must use interframe compression. This leads to the so-called "long GOP" compression artifacts. With fog, you end up with the objects seen through the fog looking just fine, but the fog itself is seen in jerks because the compression algorithms cannot track the amorphous cloud of fog, and therefore it hardly moves for 14 frames, and only when the next keyframe is stored does it reset to the proper location. So, in this case, I'd take dot crawl any day of the week, and I would say they should shoot the person who invented digital video compression. But of course, without compression, we wouldn't have digital video, and since we wouldn't have had analog video without all the things you hate, between us we'd still be waiting for television to be invented. |
|||
26th July 2016, 18:43 | #13 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
|
|
26th July 2016, 19:10 | #14 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
but compression artifacts could be removed almost completely, something that's not gonna happen on dot crawl/rainbow, especially rainbow, you filter the vid till its color looks totally washed out and you can still see some residual rainbows.. and I'd be glad to see what I can do and maybe add a new function to Oyster if you can upload a sample of that jerky vid and start another thread for that. |
|
26th July 2016, 19:57 | #15 | Link |
Registered User
Join Date: Feb 2002
Location: California
Posts: 2,691
|
If you can reduce the problem, you're a genius. I'll try to figure out how to upload enough without compromising the business negotiations. The people who own the video are trying to sell it to the car company whose vehicle is featured, and there is some real money involved. I'll start a new thread when I'm ready to do that. I want to see how far I can get by sensing the "jump" points and then using motion estimation to smooth out the video at those points.
|
26th July 2016, 20:11 | #16 | Link |
Life's clearer in 4K UHD
Join Date: Jun 2003
Location: Notts, UK
Posts: 12,219
|
Composite video was the best we had back in the mid-to-late 70's. It is what it is!
As somebody who has quite a few LaserDisc's (both NTSC and PAL), I'd be interested to see how this topic progresses. I guess the first step is obtaining a good quality analogue video to digital video capture device
__________________
| I've been testing hardware media playback devices and software A/V encoders and decoders since 2001 | My Network Layout & A/V Gear |
|
26th July 2016, 20:46 | #17 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,345
|
(sorry for thread hijack) or a mod can split this off
@feisty, while john is preparing, have a look at some of the native dji phantom 3 drone samples in this linked thread, which exhibit similar long GOP characteristics as to what I believe he is describing , but instead cloud shadows http://forum.videohelp.com/threads/3...-I-do-about-it It's not long GOP per se in general that is at fault, it's the cheap / low quality compression specifically used in the drone. You can get much higher quality at the same bitrate, with only slightly more expensive encoding hardware chip. That drone uses no b-frames . It's a cost cutting decision. So anyways there has should be a more elegant solution than "dumb" temporal smoothing, because you have a known 8frame GOP cycle, IPPPPPPP . So throw some oysters or other seafood at it Last edited by poisondeathray; 26th July 2016 at 20:49. |
26th July 2016, 22:39 | #18 | Link | |
Registered User
Join Date: Feb 2002
Location: California
Posts: 2,691
|
Quote:
And you're right: the GOP is 8 frames long. ----- [edit]Here's a link to my new thread on this subject: Long GOP Artifacts We now return you to your regular programming ... Last edited by johnmeyer; 26th July 2016 at 23:14. Reason: Added link to new thread |
|
27th July 2016, 00:30 | #19 | Link |
Registered User
Join Date: Jan 2006
Posts: 1,867
|
I just wanted to point out one case where composite is actually better from a laserdisc!
When the content is film with pulldown, sometimes a field is repeated for each phase of the dot crawl, which means if you average those fields together, you get a perfect, true s-video quality field with no dot crawl. Of course, this one perfect field isn't so useful, however I had the idea that you use it as a guide somehow to reduce dot crawl in other fields and frames, or some kind of interpolation. It seems it should work somehow, since you have this perfect reference to help you. You can't use this idea with the s-video out, I think the attempts at filtering out the dot-crawl would destroy those two fields that were in opposite phases. Also, when you average two fields, if the noise is normally distributed (a good assumption), the standard deviation of the noise is reduced by sqrt(2) or 41%. Last edited by jmac698; 27th July 2016 at 00:35. |
27th July 2016, 08:10 | #20 | Link |
Life's clearer in 4K UHD
Join Date: Jun 2003
Location: Notts, UK
Posts: 12,219
|
Hi all,
Can anyone recommend any good quality analogue video to digital video capture devices? Cheers
__________________
| I've been testing hardware media playback devices and software A/V encoders and decoders since 2001 | My Network Layout & A/V Gear |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|