VideoHelp Forum
+ Reply to Thread
Page 1 of 2
1 2 LastLast
Results 1 to 30 of 35
Thread
  1. Hello all,

    I've finally taken the plunge and invested in a JVC Super VHS (S-VHS) VCR, an EasyCap (or EZCap) 116 and have put together a workflow of sorts. I'd be really grateful if someone could take a look.

    Code:
    1. Play VHS from JVC HR-S7611 with Digital TBC (and NR) enabled, BEST on, picture control set to auto, digital 3R on.
    2. Capture using EZCap 116 via s-video and twin phono plugs for audio.
    3. Capture using virtual dub with the settings below.
    Virtual dub settings:
    Code:
    • Lagarith codec, RBG, multithreading enabled.
    • Video standard Pal I
    • Colour space YUY2
    • Frame rate 25
    • Resolution 704 x 576
    • Luma white point extended so that all data is within blue part of histogram
    • Audtio set to 48000Hz 16bit (in both vdub and windows)
    • Audio uncompressed
    I've got the following concerns:
    1. I couldn't find any reviews of the JVC HR-S7611, but it has TBC so I was happy.
    2. The EZCap 116 is similar sparse on reviews. Some people here say EZCap is good, some say throw it in the bin. I can't get hold of the ATI 600 USB, so how bad is my current setup?
    3. Is this a decent set up for my purposes? I just want to create a good quality backup, uncompressed (for now), of all my old home movies.
    4. I'm capturing at 704x576 as I've read that it's a good idea to oversample interlaced footage. Also, this resolution is exactly double tat of VHS, so I thought it would be a sensible choice when it comes to downsampling it again. (Simple skipping of every other pixel in both directions, no loss of data).

    Here's a sample:

    Click image for larger version

Name:	vlcsnap-2014-02-23-12h27m21s76.png
Views:	1157
Size:	933.8 KB
ID:	23720

    Here's a sample video:

    https://forum.videohelp.com/attachments/23723-1393167332/Jaydee%20Sample%202.avi

    I've noticed the slight banding (?) at the bottom of the screen. It gets worse if I turn the TBC off, but I can't eliminate it completely.

    So, what do you think? Am I really missing anything, or is this good for my purposes?
    Last edited by Jaydee; 23rd Feb 2014 at 09:58. Reason: added sample video
    Quote Quote  
  2. If you are planning on filtering in software disable the S-VHS decks noise reduction filter (if you can). Software filters a far superior and the deck's filter is going to mess up the video. I know that on some decks the TBC and NR are locked together so that may not be possible. When that's the case you have to decide which is the lessor evil.

    Since you are capturing YUY2 you should set Lagarith to YUY2 mode. Conversion to RGB will lose precision and can lead to loss of super-blacks and super-brights. Do not use VirtualDub to filter while capturing (it will convert to RGB).

    Capture at 704x756 or 720x576. And leave it at that. PAL VHS has an effective resolution of roughly 350x576 for the luma channel, PAL S-VHS around 480x576.

    A short video sample would be more illuminating.
    Quote Quote  
  3. I can't disable the noise reduction without disabling the TBC also, so it will have to remain I am afraid.

    Thanks for the tip on capturing in YUY2. Will do.

    Please find attached sample video.
    Image Attached Files
    Quote Quote  
  4. Originally Posted by jagabo View Post
    PAL VHS has an effective resolution of roughly 350x576 for the luma channel
    Pal vhs:
    luma: 335x576
    chroma: 40x240

    source: http://en.wikipedia.org/wiki/VHS
    *** DIGITIZING VHS / ANALOG VIDEOS SINCE 2001**** GEAR: JVC HR-S7700MS, TOSHIBA V733EF AND MORE
    Quote Quote  
  5. Member 2Bdecided's Avatar
    Join Date
    Nov 2007
    Location
    United Kingdom
    Search Comp PM
    I don't author DVDs at any of those horizontally sub-sampled resolutions. Some DVDs players do a poor job of upscaling them, and the result looks worse than the original VHS. That's even before you've tried to improve it.

    If you want to fit more than two hours onto a DVD it can be worth doing, but otherwise forget it. Stick with 704x576.

    YMMV of course - you can always try it. Just don't go ahead and do it without checking the results.


    In your sample, there's cross-colour artefacts. Are you using the S-video connection? If captured properly, it wouldn't have those fine grained diagonal artefacts in the flower and saturated colour areas. That's from chroma being interpreted as luma. That can't happen with a proper S-video connection, so something is going wrong. It's not on the tape from a previous generation of composite dubbing because the pattern is far too fine. Maybe some EZ-cap setting needs changing?

    Did you turn the TBC off? It looks like it. If so, you should have left it on.

    superwhites are clipped to y=235, whereas there are overexposed parts of the video where it obviously has values higher than this. They're all clipped to 235. You either want to make absolutely sure (using the pre-amp) that everything usable is in-range, and/or allow values higher than 235 through. I've never used an EZcap device so I can't advise on how to do this.


    The tapes seem to be playing back pretty well so hopefully when you've got these capture issues sorted you should be able to get decent results.

    Cheers,
    David.
    Quote Quote  
  6. Yes, the biggest issue is the herringbone noise. You shouldn't be getting that from an s-video signal and it's very hard to get rid of (without damaging the picture) via filtering. You should concentrate on that first.
    Quote Quote  
  7. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    There is serious crosstalk going on in the pic. If you can't get rid of it at the capture level, just resize downwards at half the horizontal resolution, and upsize back to the original resolution. If using a lossless codec, results will be better.
    Originally Posted by jagabo View Post
    Yes, the biggest issue is the herringbone noise. You shouldn't be getting that from an s-video signal and it's very hard to get rid of (without damaging the picture) via filtering. You should concentrate on that first.
    Again, it's always better to eliminate any artifacts at the capture level, and work on that first, like Jagabo says. However, IF all hope is lost (and I stress "IF"), and you capture with herringbone artifacts - there's always Neat Video for processing. With most cases, it's very easy to get rid of, with minimal damage.
    I hate VHS. I always did.
    Quote Quote  
  8. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    Ok, this insomnia won't go away unless I take a stab at this.

    Removed the crosstalk for now since it was really bugging me (via downsizing and upsizing as mentioned in my last post). No serious herringbone artifacts in my eyes other than on the red outift the woman is wearing.

    I can give details on how I simmered down the crosstalk, and details as to how to remove the herringbone noise as well, tomorrow if interested.

    Now I can rest a few winks.
    Image Attached Files
    I hate VHS. I always did.
    Quote Quote  
  9. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Yep, resizing is about the only way. That herringbone shouldn't be there in the first place. We've had several threads recently
    with VHS captured with budget cards and composite. Same problems in every one of them.

    Brights are so badly clipped, there's no way to get any dynamic range out of it. Levels change every time something moves.
    A nice, decent shot, crippled by autogain. I've said it many times, and I'll say it again: it should be illegal to sell consumer
    cameras with AGC. The best one can do here is control levels during capture to try to rescue some highlights and avoid too
    much clipping. Color tweaked in VirtualDub, but that's a matter of user preference.

    Script attached. HAd this ready yesterday, but the forum download was so slow I fell asleep! Neither PuzZLeR, myself, nor the
    O.P. should have to go through all this work. Need a better capture.
    Last edited by sanlyn; 19th Mar 2014 at 03:51.
    Quote Quote  
  10. I did this to remove the herringbone and resharpen a bit:

    Code:
    AviSource("Jaydee Sample 2.avi") 
    AssumeTFF()
    BicubicResize(320,576)
    ConvertToYV12(interlaced=true)
    #Sharpen(0.2,0.0)
    QTGMC(preset="fast")
    nnedi3_rpow2(2, cshift="Spline64Resize", fwidth=512, fheight=576)
    Sharpen(0.2,0)
    nnedi3_rpow2(2, cshift="Spline64Resize", fwidth=704, fheight=576)
    Sharpen(0.2,0.0)
    You can play around with those Sharpens if you want something sharper or less sharp. I didn't address any other issues. In fact, I didn't think this was worth addressing since he may be able to get better caps.
    Image Attached Files
    Quote Quote  
  11. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Looks good. Maybe the O.P. has questions about how to avoid some of these capture problems.

    For one thing, I can advise setting the capture device's bright6ness and levels controls to avoid wiped-out darks and brights. They can always be tweaked later. Denoisers, sharpeners, etc., are useless during VHS capture. Brightness, contrast, and maybe saturation are about all that should be used. Any other filters will get in the way.
    Last edited by sanlyn; 19th Mar 2014 at 03:51.
    Quote Quote  
  12. Member 2Bdecided's Avatar
    Join Date
    Nov 2007
    Location
    United Kingdom
    Search Comp PM
    Just to emphasise: you should capture without the herringbone effect. Find out what's wrong in the capture chain, and fix it if you can.

    It's very clever that two people can fix it after the fact, but it shouldn't be there in the first place.


    Originally Posted by sanlyn View Post
    I've said it many times, and I'll say it again: it should be illegal to sell consumer
    cameras with AGC.
    Can you imagine the results without?! People would shoot hours of almost all white and almost all black because they'd set it wrong, and couldn't tell on the poor view finder in difficult shooting conditions. Don't underestimate people's ability to get basic things wrong. They still shoot straight into the light. They still wave the camera around like a conductor's baton. What makes you think they'd get manual video gain control right?

    What helps are modern sensors with a wider dynamic range, and an ability to reduce the contrast to avoid the over-contrasty look so prevalent in consumer camcorders of the past. I've never clipped anything important with autogain on the HV20 in cinemode. I already have on the HC-X920 - it's contrast can't be sufficiently tamed . It's never going to do anything as bad as a typical 1990s consumer camcorder though.

    Consumers seems to like that overly contrasty look, and that typically means burnt out highlights and crushed blacks. It might make us cringe, but...!

    Cheers,
    David.
    Quote Quote  
  13. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Well, yes, 2BDecided, you're correct. I don't for a minute imagine that AGC will disappear from cameras, and they've certainly improved. People aren't going to get involved with the intricacies of setting their own exposure any more than they'll learn to use Avisynth and other tools for processing video. They just want to press a button or click an icon and get ye olde Instant Gratification. It makes the world go 'round, as they say.
    Last edited by sanlyn; 19th Mar 2014 at 03:52.
    Quote Quote  
  14. Member
    Join Date
    May 2007
    Location
    United States
    Search Comp PM
    Also on a side note. On the JVC's the noise reduction can be turned off by setting the system to use "edit". The TBC can be used w/o noise reduction, but noise reduction cannot be used w/o the TBC (go figure). You can also turn on/off Digital 3R (edge enhancement). Personally I leave 3R off, TBC on, and NR on edit when capturing good tapes (especially SVHS). On really noisy tapes I use NR on. D3R is good for tapes that look soft, but it does introduce artifacts (there's some threads around with examples). If someone can show me a work flow of software NR that works as good as the JVC NR you'll make me a believer .
    Quote Quote  
  15. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    Originally Posted by ybarra
    On really noisy tapes I use NR on. D3R is good for tapes that look soft, but it does introduce artifacts (there's some threads around with examples). If someone can show me a work flow of software NR that works as good as the JVC NR you'll make me a believer .
    Ever try Neat Video software? I believe those guys should be next asked to cure cancer.

    I keep DR3 on all the time, high sharpness and high details. Neat Video does the rest after the capture (as per the noise/motion artifacts/chroma dancing/low freq shifts/etc that may result). Neat Video is far, far more superior to any VCR's NR.

    You will be a believer when you try it.
    I hate VHS. I always did.
    Quote Quote  
  16. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Great stuff, but don't ever use NeatVieo's default settings. Your video will look like something made of Jello. Take it easy on the filter settings and learn to tweak its advanced interface.
    Last edited by sanlyn; 19th Mar 2014 at 03:52.
    Quote Quote  
  17. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    Originally Posted by sanlyn View Post
    Great stuff, but don't ever use NeatVieo's default settings. Your video will look like something made of Jello. Take it easy on the filter settings and learn to tweak its advanced interface.
    I agree. The default settings are horrid, and can make the video look like Jello or plastic depending on the content. Worse, the default settings may initially put off the first-time user before they realize how truly great this application is. It almost did with me during the trial.

    Take a few moments to learn how to build a proper noise profile first. There are some good video tutorials on this online (even on their site). Practice, tweak, practice, tweak on smaller clips to get the feel of it for a few days, and experiment with different settings till you feel your sweet spot, before you use it within your workflow. Soon enough you will love it and wonder how you'd function without it.

    It's not free, but very affordable. It's also quite slow in processing, but very thorough and nothing a late model PC can't handle with a few overnights. For those that don't have a faster computer right now, you can take solace in the fact that you can capture with confidence today, and clean it up down the road. You can do a few tests in the meantime for re-assurance on small clips.

    I actually know someone who bought an i7 just for Neat Video.

    Very worth it for VHS transfers - for hobbyists and professionals.
    Last edited by PuzZLeR; 28th Feb 2014 at 15:15.
    I hate VHS. I always did.
    Quote Quote  
  18. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    @Jagabo: Can I get you to rewrite your script with a few comments (#)?

    I prefer to archive interlaced, but your script looks great for playback on the more modern displays, and I'd love to tweak it for NTSC-land, and other filters I use.

    For anybody that's interested, I used a simple approach to preserve interlacing. Just downsize to half the width at 352x576, then upsize back up to 704x576. You can do this with Avisynth or VirtualDub's filters.

    For downsize, you can try bilnear or bicubic to your taste (the latter is a bit sharper).

    For upsize, you can try Lanczos(3 or 4). I prefer to use Super Resolution for upsizing. Not free, but very nice.

    http://www.infognition.com/super_resolution_vdf/
    http://www.infognition.com/super_resolution_avisynth/
    Last edited by PuzZLeR; 5th Mar 2014 at 03:20.
    I hate VHS. I always did.
    Quote Quote  
  19. Banned
    Join Date
    Oct 2004
    Location
    New York, US
    Search Comp PM
    Originally Posted by PuzZLeR View Post
    For anybody that's interested, I used a simple approach to preserve interlacing. Just downsize to half the width at 352x576, then upsize back up to 704x576. You can do this with Avisynth or VirtualDub's filters.

    For downsize, you can try bilnear or bicubic to your taste (the latter is a bit sharper).

    For upsize, you can try Lanczos(3 or 4). I prefer to use Super Resolution for upsizing. Not free, but very nice.


    Never mind. Maybe it's too early in the day and I just didn't read this correctly.
    Last edited by sanlyn; 19th Mar 2014 at 03:52.
    Quote Quote  
  20. Member 2Bdecided's Avatar
    Join Date
    Nov 2007
    Location
    United Kingdom
    Search Comp PM
    Originally Posted by PuzZLeR View Post
    I prefer to archive interlaced
    ... then you'd just re-interlace it at the end.

    However, when ever I perform some restoration processes that requires (or works better after) deinterlacing (bobbing), I always archive the deinterlaced result - unless there was something about the restoration that meant I didn't use the best deinterlacing.

    e.g. If I used a dumb bob(), because that was more than good enough for my purposes, and always intended to output interlaced preserving the original fields, then the interlaced output is what I'd archive.

    However, if I'd used QTGMC in a half-decent mode (or better), then I would keep the progressive result. It's trivial to re-interlace it if I want interlaced later, but if I'd archived interlaced I'd have to put it through QTGMC a second time if I want progressive later, and that's undesirable. It's a waste of time, and running QTGMC twice just gets you further away from the original video (unless you use preserve original fields, which IME doesn't give you as good deinterlacing).

    Cheers,
    David.
    Quote Quote  
  21. Originally Posted by PuzZLeR View Post
    @Jagabo: Can I get you to rewrite your script with a few comments (#)?
    I thought it was pretty straight forward but...

    Code:
    # read an interlaced PAL source, 704x576
    AviSource("Jaydee Sample 2.avi") 
    
    # make sure the field order is flagged properly
    AssumeTFF()
    
    # dot crawl filters like CheckMate() and DrCrawl() work well
    # with the two frame repeating dot pattern of NTSC video.
    # But they don't work well with the 4 frame repeating pattern
    # of PAL video.  Downsizing the video horizontally will blur away
    # the dot crawl / herringbone noise of both PAL and NTSC. 
    # This doesn't hurt the resolution of VHS sources too much
    # because VHS only has about 350 lines of resolution.
    # You can play around with the horizontal size here to fine tune
    # for your particular video.  Smaller sizes will blur away more of
    # the noise but blur the underlying picture more.  Larger
    # sizes may be sufficient for less severe noise and
    # will retain a sharper (if your source has it) picture.
    # Do not resize vertically here, it will screw up the interlacing.
    BicubicResize(320,576)
    
    # QTGMC() will require YV12
    ConvertToYV12(interlaced=true)
    
    # Sharpen a bit at the current low frame size if necessary.
    # If the video is already sharp this will generate halos.
    #Sharpen(0.2,0.0)
    
    # smart bob deinterlace and some edge cleanup
    # you can use Yadif() or some other smart bob
    # here if you need faster processing or don't
    # need the higher quality of QTGMC().
    # Use whatever settings you want here.
    QTGMC(preset="fast")
    
    # upscale ~half way to our desired final size
    # nnedi3_rpow2() produces less aliasing than
    # simple resizers like LanczosResize()
    nnedi3_rpow2(2, cshift="Spline64Resize", fwidth=512, fheight=576)
    
    # and sharpen the result a bit, only horizontal sharpen
    # (unless your source needs vertical sharpening too)
    Sharpen(0.2,0)
    
    # upscale to our desired final size
    nnedi3_rpow2(2, cshift="Spline64Resize", fwidth=704, fheight=576)
    
    # and sharpen a bit more
    Sharpen(0.2,0.0)
    
    # The above sequence using nnedi3/shapren twice produces a sharper
    # picture than simply upscaling once to the desired size and sharpening.
    # You can also use multiple nned3_rpow2().Sharpen() calls to upscale
    # to high def resolutions -- which can work very well with cartoons.
    # you might throw a mild aWarpSharp2() or two into the sequence
    # for additional edge sharpening.
    
    # if you need interlaced output you can re-interlace the video here:
    AssumeTFF() # or BFF, whichever you prefer
    SeparateFields() # 50 frames per second to 100 fields per second
    SelectEvery(4,0,3) # get rid of the two middle fields of every 4, 50 fields per second
    Weave() # weave remaining fields together 25 fps, interlaced
    I didn't test this script after adding all the annotation so you should check it...
    Last edited by jagabo; 5th Mar 2014 at 07:49.
    Quote Quote  
  22. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    @Jagabo: I was curious about a couple of things (like the double upsize, and the sequence, etc.), and this clears things. Your noting was more thorough than I bargained for - thanks!

    @2BDecided: Maybe it's my eyes, but interlaced video on certain displays (that display content deinterlaced at the display level better than content that was deinterlaced at the encoding level) looks so much better.

    We are indeed entering an age though where filters like QTGMC() give us more confidence to archive deinterlaced. However, unless I missed something, wouldn't you need to archive it in a double frame-rate to properly restore interlace down the road? That's like twice the file size. (But I should test this more.)

    Originally Posted by sanlyn


    Never mind. Maybe it's too early in the day and I just didn't read this correctly.
    All I was doing was demonstrating some methods to effectively remove the dotcrawl/crosstalk with very simple filters, and relatively quick processing time. I figure post-processing addicts like us would appreciate that.
    I hate VHS. I always did.
    Quote Quote  
  23. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    @Jaydee: If still alive, I'm still curious how you get that dot crawl pattern.

    Although the ezcap has no comb filter (or the ATI 600 USB if wishing to get one), your JVC does have one. Sometimes this can be caused by an external TBC, but since you don't have one in the chain, and you're using s-video, this is all moot. I'd say try different settings in your JVC (I prefer the EDIT option as well, and play around with NR/TBC and the Video Stabilizer).

    I only get such artifacts when I have no choice but to use a composite/RCA output only VCR (when my s-video ones DO NOT BEHAVE on a certain tape ).

    It should be better to try to clear this at the capture level, however putting a DVR in the chain to reduce dot crawl kills quality IMO. Since it's easily corrected using a lossless format, it shouldn't be a big deal.

    *NOTE: If anyone is thinking that a JVC S-VHS VCR would behave as a comb filter in passthrough, it WON'T work. It's only effective at the unit's tape playing level.
    Last edited by PuzZLeR; 5th Mar 2014 at 12:07.
    I hate VHS. I always did.
    Quote Quote  
  24. Originally Posted by PuzZLeR View Post
    It should be better to try to clear this at the capture level,
    Which is why I mentioned he shouldn't be getting this herringbone noise from an s-video source and he needs to look into that first.
    Quote Quote  
  25. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    Originally Posted by jagabo View Post
    Yes, the biggest issue is the herringbone noise. You shouldn't be getting that from an s-video signal and it's very hard to get rid of (without damaging the picture) via filtering. You should concentrate on that first.
    Originally Posted by jagabo View Post
    Originally Posted by PuzZLeR View Post
    It should be better to try to clear this at the capture level,
    Which is why I mentioned he shouldn't be getting this herringbone noise from an s-video source and he needs to look into that first.
    Yes indeed. It's just that I went slightly off topic with filtering, so wanted to redeem the thread.

    But, as an aside, is crosstalk/dot crawl synonymous with "herringbone noise"? I thought they were different artifacts, the latter being wavy lines, such as those you may notice on a recording of a bad channel. (I noticed some on the woman's red dress in the clip.) Didn't mean to confuse if so.

    At any rate, I'm starting to think a bad cable is a possibility here.
    I hate VHS. I always did.
    Quote Quote  
  26. Originally Posted by PuzZLeR View Post
    But, as an aside, is crosstalk/dot crawl synonymous with "herringbone noise"?
    In this case it is.

    Originally Posted by PuzZLeR View Post
    I thought they were different artifacts, the latter being wavy lines, such as those you may notice on a recording of a bad channel.
    Yes, herringbone is a wider class that describes any noise that shows up as diagonal or zig-zag lines.
    Quote Quote  
  27. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    Thanks Jagabo.

    As for the pattern that is definitely in the crosstalk/dot crawl subset, which looks obviously like Y/C leakage, I've become content to just filter it out. Using lossless formats, as well as good software solutions (and a good processor) has made a difference.

    At least I know what's causing it on my end on those specific composite/RCA captures.

    The reason that I haven't bothered too much adding in a comb filter solution to the chain is it is also a pain. Capture devices, such as the ATI 650 and ATI 750 have comb filters, but they are plagued with AGC issues. Worse, the 750 is a bad product IMO for other reasons too.

    Some DVRs have an effective passthrough, but weaken quality. I believe, for those that "work", they recompress/reencode the video along the way, which create other obvious artifacts.

    In this case, or even in the case of the O/P if still wishing to seek a solution at the capture level when the current problem can't be detected, is there anything else anyone can recommend? On my end it's hypothetical, since I've considered it resolved years ago, but I'm just curious if anyone knows of any other solution worth trying today.
    I hate VHS. I always did.
    Quote Quote  
  28. Member 2Bdecided's Avatar
    Join Date
    Nov 2007
    Location
    United Kingdom
    Search Comp PM
    Originally Posted by PuzZLeR View Post
    @2BDecided: Maybe it's my eyes, but interlaced video on certain displays (that display content deinterlaced at the display level better than content that was deinterlaced at the encoding level) looks so much better.
    I can't say I have a 100% fair way of comparing like-for-like when one variable is the TV's deinterlacing. I do know that de-interlacing, re-interlacing, and then de-interlacing again looks worse on a PC. I also know that the deinterlacing in modern TVs is hugely variable. Sometimes it can be subjectively better than QTGMC though.

    We are indeed entering an age though where filters like QTGMC() give us more confidence to archive deinterlaced. However, unless I missed something, wouldn't you need to archive it in a double frame-rate to properly restore interlace down the road? That's like twice the file size. (But I should test this more.)
    For many lossless formats, yet, twice the size. There's a lot of redundancy in there though which can be exploited.

    I'm not saying deinterlace for archiving. I'm saying that, if you've already deinterlaced properly as part of the restoration process, it's pointless to re-interlace before archiving. I would never ever ever convert true 50i to 25p for archiving. It's got to be 50p or 50i.


    Originally Posted by PuzZLeR View Post
    @Jaydee: If still alive, I'm still curious how you get that dot crawl pattern.

    Although the ezcap has no comb filter (or the ATI 600 USB if wishing to get one), your JVC does have one.
    Not in this context. Luma and Chroma are stored separately on VHS and S-VHS. The VCR doesn't have to separate them during playback.

    Cheers,
    David.
    Quote Quote  
  29. Member PuzZLeR's Avatar
    Join Date
    Oct 2006
    Location
    Toronto Canada
    Search Comp PM
    Originally Posted by 2Bdecided
    I can't say I have a 100% fair way of comparing like-for-like when one variable is the TV's deinterlacing. I do know that de-interlacing, re-interlacing, and then de-interlacing again looks worse on a PC. I also know that the deinterlacing in modern TVs is hugely variable. Sometimes it can be subjectively better than QTGMC though.
    It is indeed more subjective than anything, and it would be difficult to compare. Yes, on modern displays anything can happen but likely can be unforgiving. But on older SD TVs, or even with Windows Media Player's default de-interlacer during playback, the DvDs/MPEG-2 streams (after encoding from lossless) just play more smoothly in motion, and crisper in picture quality - with the interlace intact. I know there are some sharpening filters under the hood during playback, but they wouldn't have the same effect with deinterlaced video. Again, it's just my eyes.

    Originally Posted by 2Bdecided
    For many lossless formats, yet, twice the size. There's a lot of redundancy in there though which can be exploited.

    I'm not saying deinterlace for archiving. I'm saying that, if you've already deinterlaced properly as part of the restoration process, it's pointless to re-interlace before archiving. I would never ever ever convert true 50i to 25p for archiving. It's got to be 50p or 50i.
    Yes, I agree it's wasteful to re-interlace after such a step. But I do do my best to keep interlacing in the archiving whereever possible to have more choices.

    This is one reason why I asked Jagabo to comment on his script, which deinterlaces. I wanted to actually understand it a bit better, so I can play around with it and keep the interlacing. (Don't tell him. )

    Most restoration processes I use happily keep interlacing. For some, it would be a disaster, which, yes I agree, would be ideal to archive with the interlacing removed.

    Originally Posted by 2Bdecided
    Not in this context. Luma and Chroma are stored separately on VHS and S-VHS. The VCR doesn't have to separate them during playback.
    Ah, Ok. What is strange though about this process is that the external TBC I use, the AVT-8710, doesn't separate luma/chroma properly from the composite decks I use, unless I'm doing something wrong. However, when inputing a composite signal from the JVC S-VHS deck, there's no problem, or I'm doing something right in this case.

    Go figure. This is why I thought something was done internally in the JVC during playback. (But, like I said, it won't work as a passthrough.)
    I hate VHS. I always did.
    Quote Quote  
  30. Member 2Bdecided's Avatar
    Join Date
    Nov 2007
    Location
    United Kingdom
    Search Comp PM
    Originally Posted by PuzZLeR View Post
    Originally Posted by 2Bdecided
    I can't say I have a 100% fair way of comparing like-for-like when one variable is the TV's deinterlacing. I do know that de-interlacing, re-interlacing, and then de-interlacing again looks worse on a PC. I also know that the deinterlacing in modern TVs is hugely variable. Sometimes it can be subjectively better than QTGMC though.
    It is indeed more subjective than anything, and it would be difficult to compare. Yes, on modern displays anything can happen but likely can be unforgiving. But on older SD TVs, or even with Windows Media Player's default de-interlacer during playback, the DvDs/MPEG-2 streams (after encoding from lossless) just play more smoothly in motion, and crisper in picture quality - with the interlace intact.
    Sorry, maybe we're talking at cross purposes here. SDTVs can't play progressive material - not 50p or 60p anyway. That would be ED. DVDs certainly don't support it. If you're comparing 50i with 25p (or 60i with 30p), then of course 25p is lacking. It's lacking half the temporal resolution.

    Originally Posted by 2Bdecided
    Not in this context. Luma and Chroma are stored separately on VHS and S-VHS. The VCR doesn't have to separate them during playback.
    Ah, Ok. What is strange though about this process is that the external TBC I use, the AVT-8710, doesn't separate luma/chroma properly from the composite decks I use
    ...but if you're using a composite output, the separate luma and chroma have been put (back) together for the composite output, so are no longer separate. Though they don't overlap with VHS sources because the luma stops around 3MHz. See...
    http://commons.wikimedia.org/wiki/File:S-video_spectrum.png
    ...a) is composite, b) is S-video, and VHS doesn't have any Y/luma above 3MHz so in theory doesn't need S-video. In practice it's safer to use S-video in case other devices don't do the filtering of composite properly (even though the filtering for a VHS source is trivial due to the lack of overlap).

    For all the theory, I'm a great believer in "whatever works" - with the warning that sometimes things look like they're working, but are causing other problems which you notice later.

    Cheers,
    David.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!