VideoHelp Forum
+ Reply to Thread
Page 2 of 3
FirstFirst 1 2 3 LastLast
Results 31 to 60 of 78
Thread
  1. Originally Posted by energizerfellow View Post
    Originally Posted by just4747 View Post
    Unfortunately my FireWire PCI-E card completely stopped working and only blue screens my PC now all of a sudden, so I am waiting for another one to come tomorrow.
    Make sure it's a IEEE 1394b card based on the Texas Instruments XIO2213B as it's basically the only readily available PCIe chipset that's proven to be reliable. The IEEE 1394a cards based on VIA chipsets are hot garbage.
    AFAIK, the difference between "a" and "b" is the max speed (400 vs 800 Mbit/s) and the connector shape (6-pin vs 9-pin). If someone has a 6-pin cable, is it prudent to buy a 9-pin card? As for VIA being "hot garbage", I just bought a VIA-based FireWire card, but haven't installed it. I'll report back how it works.
    Quote Quote  
  2. Originally Posted by ConsumerDV View Post
    AFAIK, the difference between "a" and "b" is the max speed (400 vs 800 Mbit/s) and the connector shape (6-pin vs 9-pin). If someone has a 6-pin cable, is it prudent to buy a 9-pin card? As for VIA being "hot garbage", I just bought a VIA-based FireWire card, but haven't installed it. I'll report back how it works.
    The differences between FireWire 400 (IEEE 1394a-2000) and FireWire 800 (IEEE 1394b-2002) are quite extensive and while you may not need the speed, you'll be using a more modern host controller chipset.

    The IEEE 1394b cards usually have both a 6-pin and 9-pin connector on them. I generally like to use one-piece cable with whatever ends are needed to avoid things like loose connections and insertion loss.
    Quote Quote  
  3. Originally Posted by energizerfellow View Post
    The differences between FireWire 400 (IEEE 1394a-2000) and FireWire 800 (IEEE 1394b-2002) are quite extensive and while you may not need the speed, you'll be using a more modern host controller chipset.

    The IEEE 1394b cards usually have both a 6-pin and 9-pin connector on them. I generally like to use one-piece cable with whatever ends are needed to avoid things like loose connections and insertion loss.
    Thanks, it is a nice presentation. It seems that the major differences between "a" and "b" are longer cables including fiber optical connections, and arbitration in multi-node networks. This is not important for my use case of a single device connected to a host with a 1-meter cable. So, I hope that VIA-based "hot garbage" works for me
    Quote Quote  
  4. Member Cornucopia's Avatar
    Join Date
    Oct 2001
    Location
    Deep in the Heart of Texas
    Search PM
    Nothing at all wrong with 1394a as that was the spec that dv was designed to use. The chipset, OTOH, greatly matters, and time has over and again shown to substantiate the claim of sticking with TI.


    Scott
    Quote Quote  
  5. Member
    Join Date
    Feb 2008
    Location
    Poland
    Search Comp PM
    Originally Posted by ConsumerDV View Post
    AFAIK, the difference between "a" and "b" is the max speed (400 vs 800 Mbit/s) and the connector shape (6-pin vs 9-pin). If someone has a 6-pin cable, is it prudent to buy a 9-pin card?
    There is completely no diffrence in features between A and B.
    Quote Quote  
  6. Member pchan's Avatar
    Join Date
    Mar 2003
    Location
    Singapore
    Search Comp PM
    Originally Posted by ConsumerDV View Post
    Originally Posted by energizerfellow View Post
    Originally Posted by just4747 View Post
    Unfortunately my FireWire PCI-E card completely stopped working and only blue screens my PC now all of a sudden, so I am waiting for another one to come tomorrow.
    Make sure it's a IEEE 1394b card based on the Texas Instruments XIO2213B as it's basically the only readily available PCIe chipset that's proven to be reliable. The IEEE 1394a cards based on VIA chipsets are hot garbage.
    AFAIK, the difference between "a" and "b" is the max speed (400 vs 800 Mbit/s) and the connector shape (6-pin vs 9-pin). If someone has a 6-pin cable, is it prudent to buy a 9-pin card? As for VIA being "hot garbage", I just bought a VIA-based FireWire card, but haven't installed it. I'll report back how it works.
    I used to using a NEC firewire card until it stopped working recently. I have to used the VIA firewire card that I put aside for a long time.
    The driver from here works. Need to put everything in a folder and do the driver update from Device Manager.
    https://www.driverscape.com/download/via-1394-ohci-compliant-host-controller
    Last edited by pchan; 6th May 2022 at 21:26.
    Quote Quote  
  7. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by ConsumerDV View Post
    DV route is straighforward, predictable, reliable and provides known quality without fiddling with various settings.
    This simply is not true.

    DV is 1990s technology, because hard drives and CPUs were too slow for better quality. Those were the days of Pentium II and III computers, the minimum specs of the DV boxes.

    Lossless and MPEG were both possible with Pentium 4 in the early 2000s.

    There's really nothing more, or less, fiddly about either. There are cheap crappy DV boxes, and cheap crappy lossless and MPEG cards. There's really no "good" DV boxes, but the Canopus are certainly passable if you want lower quality output (NTSC bad, PAL passably decent). The variety of lossless and MPEG cards is wide. But at this late date, we tend to know what's what.

    Comparing a Easycap (Easycrap) Chinese USB cards, to a Canopus DV box, is a distorted comparison. There are better options compared to both of those.
    Want my help? Ask here! (not via PM!)
    FAQs: Best Blank DiscsBest TBCsBest VCRs for captureRestore VHS
    Quote Quote  
  8. Formerly 'vaporeon800' Brad's Avatar
    Join Date
    Apr 2001
    Location
    Vancouver, Canada
    Search PM
    Originally Posted by lordsmurf View Post
    There's really nothing more, or less, fiddly about either. There are cheap crappy DV boxes, and cheap crappy lossless and MPEG cards. There's really no "good" DV boxes, but the Canopus are certainly passable if you want lower quality output (NTSC bad, PAL passably decent). The variety of lossless and MPEG cards is wide. But at this late date, we tend to know what's what.

    Comparing a Easycap (Easycrap) Chinese USB cards, to a Canopus DV box, is a distorted comparison. There are better options compared to both of those.
    What does Canopus have to do with this?

    The context of the post you quoted is DV conversion using Digital8 camcorder, which is also the context of this entire thread. I don't think ConsumerDV has ever advocated Canopus boxes, or possibly even ever used one. I've watched all of his YouTube videos about digitization, and the name has never come up.

    Popping a Video8 tape into a Digital8 camcorder and spitting out a file using WinDV is way less fiddly and way more consistent than the array of things that can go wrong for a newbie attempting lossless capture. Yes, this doesn't change the fact that excellent lossless capture is going to exceed the best DV conversion possible. Using Digital8 as a direct player-converter or a passthrough produces a pretty consistent result. Using lossless can swing from way worse than that (as seen here) or way better (as you and I strive for).


    ---

    OP just4747 join date May 4, last login the next day. I wonder if they will be back. I want more clips of this unicorn Video8 tape that looks too good to be true and which camcorder it was originally recorded on.
    Last edited by Brad; 7th May 2022 at 14:40.
    My YouTube channel with little clips: vhs-decode, comparing TBC, etc.
    Quote Quote  
  9. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by Brad View Post
    What does Canopus have to do with this?
    Not much. This is what happens when you have multiple tabs open. Doh!

    However...
    Popping a Video8 tape into a Digital8 camcorder and spitting out a file using WinDV is way less fiddly and way more consistent than the array of things that can go wrong for a newbie attempting lossless capture.
    Maybe, maybe not.

    - WinDV is still very finicky and fiddly.
    - You still have the lost footage issue to contend with, using DV camcorder transfer. Sometimes those seconds matter.
    - Hi8 loses quality played through almost all D8 (DV) camcorders, as you mention.

    Video8 and Hi8 are also a drop-happy formats. The amount of dropped frames can be insane at times. Mere line TBC in the camera won't fix that. And there's nothing magical about DV inherently that prevents dropped frames (contrary to some rubbish found online, believed by gullible newbies) -- and no, nothing you have said, just adding the the conversation here. It's BS, and it's out there.

    You (again, not you Brad, generic "you") can get away for D8 camera for Hi8 tapes, but usually only for a % of the tapes, not all. You'll still have to contend with the failed tapes, and you'll be right back here again asking for more help. I've seen that loop play out literally for decades now.
    Want my help? Ask here! (not via PM!)
    FAQs: Best Blank DiscsBest TBCsBest VCRs for captureRestore VHS
    Quote Quote  
  10. Capturing Memories dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Member Since 2005, Re-joined in 2016
    Search PM
    Firewire connection can be as frustrating as USB if one doesn't know how to troubleshoot it or never dealt with its drivers, But I suppose if both ports work up to the driver level, USB can still have problems in capturing while firewire should stream the contents with no problems. It's no doubt that capturing lossless is far superior than capturing to a codec regardless what port is used, All ports can capture lossless with special hardware and drivers, USB, Firewire, PCI, SDI, thunderbolt.
    Quote Quote  
  11. Originally Posted by Skiller View Post
    Message
    Originally Posted by ConsumerDV View Post
    Message
    Originally Posted by energizerfellow View Post
    Message
    Originally Posted by lordsmurf View Post
    Message
    Originally Posted by Brad View Post
    Message
    Just quoting names so that you get notifications to come back hopefully, lol

    Ooook..... thanks everyone so far. A few updates in reply to several of the latest comments here:

    1. As I mentioned, my previous VIA Firewire card suddenly stopped working so today I just got this TI-based 1394B PCIe Firewire card instead with the XIO2213B chipset, and an appropriate 9-to-4-pin cable for the B standard, so that should be what energizerfellow suggested. It's plugged in and working well.

    2. Since someone asked, I went back and checked the settings for the original DV capture from VDub that I posted, and under compression/codec settings in VDub, it simply said "No recompression: dvsd" and "Delta FOURCC code". In VLC that file shows as format Planar 4:1:1 YUV and 29.97 framerate. Does all of this line up with straight DV capture and no compression? Just to confirm. And if so, should the file size really have been as large as 200MB for a 1 minute clip?

    3. So now, testing the new Firewire card and trying to capture with both WinDV and Scenalyzer (trial), I am getting almost identical results with both (and also almost identical to the DV capture from VDub with the previous FW card) and I am confused again going off of what people have been saying I should expect using these. Neither program gives any settings/options for capture as far as compression/codecs, format, resolution, etc... so I assume they are both just meant for straight uncompressed, simple capturing (right?). That's fine and makes sense.

      With both programs, I ended up with almost identally sized AVI files for the same 1 minute clip as I originally posted - both about 217MB (slightly larger than the VDub DB file I had). Both say in VLC the following for info:
      Codec: DV Video (dvsd)
      Video Res: 720x480
      Buffer dim: 736x480 (why is this again automatically different than the resolution??)
      Framerate: 29.97
      Decoded format: Planar 4:1:1 YUV


    4. Is the above what I should be expecting for the details and the file size, and what I should be wanting using this setup and these programs? And if this is all good, won't my up to 2-hour videos be absolutely humongous if I capture them all this way? And is it worth it then vs a different way/not direct uncompressed capture?

    5. Brad: Of course I am back, I want to figure this sh*t out and start capturing lol. Just had to wait for my replacement FW card. And you make it sound like I am lying about the tape format and the original video (??). Like I said and posted pics of, this was an analog Video8 tape and I am not sure what camera it was originally recorded on by my Dad but the capturing camera I am using is a Sony Digital8 DCR-TRV310 that I just got from eBay last week. Is there a way I can find out which camera recorded video onto the tapes? (Doubt it). All of the other tapes I have are also Video8 tapes. Why is this hard to believe? Is the quality of my DV clip really better looking than a Video8 captured through a Digital8 camera should be?? I have 0 reason to make any of these details up btw.

    6. I have read numerous times on several forums again that the best method to capture the highest quality from an analog Video8 or Hi8 tape should be not capturing this way but with something like VirtualDub and a lossless codec like HuffYUV or Lagarith. I literally read the quote: "You won't get 'highest quality' capturing analog source to DV. It's best captured to lossless YUY2 AVI using lossless compressors like huffyuv or Lagarith."

      And while I did originally try this and was unable to use the codecs or the YUV2 4:2:2 format suggested (VDub always gave me an error about the source video format, blah blah), I am still uncertain whether I should be continuing to try going that route and figuring it out or if the simpler capture with WinDV is the way to go like most of you are saying. Which one will in fact be better output with the hardware I have available? I keep reading differing opinions.


    Thanks again all! Glad to almost be ready to have a workflow to start capturing (these 40) tapes finally. If it helps for me to post the new 1-minute clips from the WinDV and Scenalyzer, please let me know so that I can confirm they are the results I should be getting.
    Last edited by just4747; 7th May 2022 at 17:41.
    Quote Quote  
  12. Formerly 'vaporeon800' Brad's Avatar
    Join Date
    Apr 2001
    Location
    Vancouver, Canada
    Search PM
    Originally Posted by just4747 View Post
    Just quoting names so that you get notifications to come back hopefully, lol
    Nah, the forum software doesn't work like Twitter or Discord. No @ notification or similar.

    under compression/codec settings in VDub, it simply said "No recompression: dvsd" and "Delta FOURCC code". In VLC that file shows as format Planar 4:1:1 YUV and 29.97 framerate. Does all of this line up with straight DV capture and no compression? Just to confirm. And if so, should the file size really have been as large as 200MB for a 1 minute clip?
    Yes and yes. 25Mbps video + 1.5Mbps audio = ~3.3MB/sec = 199MB/min.

    With both programs, I ended up with almost identally sized AVI files for the same 1 minute clip as I originally posted - both about 217MB (slightly larger than the VDub DB file I had).
    Possibly the difference between Type-1 and Type-2 DV capture? IMO using Type-1 is preferable.

    Is the above what I should be expecting for the details and the file size, and what I should be wanting using this setup and these programs? And if this is all good, won't my up to 2-hour videos be absolutely humongous if I capture them all this way?
    It's about 13GB/hr for DV capture, which isn't considered that large for video these days. "Proper" (lossless) capture via S-Video would produce files more than double that size; ~30GB/hr but unlike DV the bitrate isn't guaranteed.

    This is the capture step. Your goal at this stage isn't to produce the final viewable files ("delivery format" with higher compression and typically 4:2:0 subsampling) because doing that in real-time results in lower quality than capturing first and encoding after.

    Brad: ... you make it sound like I am lying about the tape format and the original video (??). [...] Why is this hard to believe? Is the quality of my DV clip really better looking than a Video8 captured through a Digital8 camera should be?? I have 0 reason to make any of these details up btw.
    My apologies, that wasn't my intention. I'm saying your Video8 capture looks better than any I have ever managed. In fact it looks more natural than any Hi8 capture I have done, too.

    Is there a way I can find out which camera recorded video onto the tapes? (Doubt it).
    You're correct.

    Which one will in fact be better output with the hardware I have available? I keep reading differing opinions.
    You probably need to replace your cheap USB capture stick with something better if you want to pursue the higher-quality lossless path.

    If it helps for me to post the new 1-minute clips from the WinDV and Scenalyzer, please let me know so that I can confirm they are the results I should be getting.
    I'm just interested in seeing more clips from other sections of this tape, and others, for the reasons I stated above.
    Last edited by Brad; 7th May 2022 at 19:22.
    My YouTube channel with little clips: vhs-decode, comparing TBC, etc.
    Quote Quote  
  13. Capturing Memories dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Member Since 2005, Re-joined in 2016
    Search PM
    I've had a chance to use full size consumer VHS as well as prosumer Video8 camcorders to shoot wedding and graduation events back in the day and the quality was way better than the average mom and dad family camcorder, The imaging sensor on those was very good, the tape itself was the bottle neck for quality.
    Quote Quote  
  14. Originally Posted by just4747 View Post
    Is there a way I can find out which camera recorded video onto the tapes? (Doubt it).
    Usually the camera which recorded the video has the least tracking problems (dropouts, streaks, garbled frames ...) for playing the tape. So this may be an indication.

    And while I did originally try this and was unable to use the codecs or the YUV2 4:2:2 format suggested (VDub always gave me an error about the source video format, blah blah), I am still uncertain whether I should be continuing to try going that route and figuring it out or if the simpler capture with WinDV is the way to go like most of you are saying. Which one will in fact be better output with the hardware I have available? I keep reading differing opinions.
    If you plan to experiment with the S-video route you should first install a codec (huffyuv, lagarith, magicyuv) which works properly under VDub or AmarecTV. And you may have to replace your capturing device as has been said by others.
    If the final file size really matters to you, you would have to re-encode (=recompress) the capture in post processing. A good S-video 4:2:2 capture would be the preferred "master" for a subsequent re-encode.
    You will have to take the learning curve, and I highly recommend to use a less precious/less important tape for trying S-video capturing as you may have to restart, rewind, redo etc. dozens of times which will put extra wear and stress on the precious tapes.
    Quote Quote  
  15. Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    just quoting names so that you get notifications to come back hopefully, lol
    nah, the forum software doesn't work like twitter or discord. No @ notification or similar.
    that's annoying lol

    Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    under compression/codec settings in vdub, it simply said "no recompression: Dvsd" and "delta fourcc code". In vlc that file shows as format planar 4:1:1 yuv and 29.97 framerate. Does all of this line up with straight dv capture and no compression? Just to confirm. And if so, should the file size really have been as large as 200mb for a 1 minute clip?
    yes and yes. 25mbps video + 1.5mbps audio = ~3.3mb/sec = 199mb/min.
    Ok good to know, thanks

    Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    with both programs, i ended up with almost identally sized avi files for the same 1 minute clip as i originally posted - both about 217mb (slightly larger than the vdub db file i had).
    possibly the difference between type-1 and type-2 dv capture? imo using type-1 is preferable.
    Ok - and I've been using type 1 in WinDV so good to know.

    Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    is the above what i should be expecting for the details and the file size, and what i should be wanting using this setup and these programs? And if this is all good, won't my up to 2-hour videos be absolutely humongous if i capture them all this way?
    it's about 13gb/hr for dv capture, which isn't considered that large for video these days. "proper" (lossless) capture via s-video would produce files more than double that size; ~30gb/hr but unlike dv the bitrate isn't guaranteed.

    This is the capture step. Your goal at this stage isn't to produce the final viewable files ("delivery format" with higher compression and typically 4:2:0 subsampling) because doing that in real-time results in lower quality than capturing first and encoding after.
    Makes sense. If all I want to do with the videos after DV capturing is to make slight edits (cutting, trimming, joining or removing scenes, titles, color correction, etc.. in Premier Pro, where can I find info on how I should be setting up the sequences in Premier and doing the exports as far as video settings go to get the best quality I can in the final files with more reasonable file sizes? In the end these will mostly be viewed either on a PC with VLC or other player or via YouTube/other video site by family members.

    Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    brad: ... You make it sound like i am lying about the tape format and the original video (??). [...] why is this hard to believe? Is the quality of my dv clip really better looking than a video8 captured through a digital8 camera should be?? I have 0 reason to make any of these details up btw.
    my apologies, that wasn't my intention. I'm saying your video8 capture looks better than any i have ever managed. In fact it looks more natural than any hi8 capture i have done, too.
    Ok sorry, my bad. Good to know it looks so good I guess lol. Definitely more than good enough quality for my purposes of storing and re-watching with family too.

    Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    is there a way i can find out which camera recorded video onto the tapes? (doubt it).
    you're correct.
    Ok thx.

    Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    which one will in fact be better output with the hardware i have available? I keep reading differing opinions.
    you probably need to replace your cheap usb capture stick with something better if you want to pursue the higher-quality lossless path.
    I was referring more to what I am doing/using now vs using that same setup but with the often recommended VirtualDub/HuffYUV. I keep reading people recommending to use those for better/lossless capture with the same DV/Firewire setup I am using now. Is it worth it or should I just stick with WinDV? I've read countless times that doing DV transfer the way I am now via just WinDV (lossy) is not as good vs using VDub/HuffyUV which will be lossless.

    Originally Posted by brad View Post
    Originally Posted by just4747 View Post
    if it helps for me to post the new 1-minute clips from the windv and scenalyzer, please let me know so that i can confirm they are the results i should be getting.
    i'm just interested in seeing more clips from other sections of this tape, and others, for the reasons i stated above.
    I will def put together some more clips from other vids to show you - thanks.


    And one more question if you don't mind - I mentioned that the files I am getting via WinDV now are still saying 720x480 for resolution but 736x480 for buffer dimensions. Is this normal or is this maybe not accurate because VLC is guessing? Since I have no control over what settings it records with via this method, should I be trying to change anything or is this normal/expected? And aside from the weird 736 number, why exactly is it exporting at 720x480 resolution? Isn't Video8 by default 640x480 originally? Or does it have to do with the fact that I am digitizing using a Digital8 camera and Firewire? Thanks.
    Last edited by just4747; 9th May 2022 at 15:37.
    Quote Quote  
  16. Originally Posted by Sharc View Post
    Originally Posted by just4747 View Post
    Is there a way I can find out which camera recorded video onto the tapes? (Doubt it).
    Usually the camera which recorded the video has the least tracking problems (dropouts, streaks, garbled frames ...) for playing the tape. So this may be an indication.

    And while I did originally try this and was unable to use the codecs or the YUV2 4:2:2 format suggested (VDub always gave me an error about the source video format, blah blah), I am still uncertain whether I should be continuing to try going that route and figuring it out or if the simpler capture with WinDV is the way to go like most of you are saying. Which one will in fact be better output with the hardware I have available? I keep reading differing opinions.
    If you plan to experiment with the S-video route you should first install a codec (huffyuv, lagarith, magicyuv) which works properly under VDub or AmarecTV. And you may have to replace your capturing device as has been said by others.
    If the final file size really matters to you, you would have to re-encode (=recompress) the capture in post processing. A good S-video 4:2:2 capture would be the preferred "master" for a subsequent re-encode.
    You will have to take the learning curve, and I highly recommend to use a less precious/less important tape for trying S-video capturing as you may have to restart, rewind, redo etc. dozens of times which will put extra wear and stress on the precious tapes.
    Probably not interested in using S-Video at this point but was mainly wondering why so many people suggest using the same method I am now (DV via a camera) but with VDub/HuffyUV vs WinDV like I am currently. Is the lossless codec and VDub really worth it? How different will it be vs just using WinDV in terms of quality file size?
    Quote Quote  
  17. Originally Posted by just4747 View Post
    why exactly is it exporting at 720x480 resolution? Isn't Video8 by default 640x480 originally? Or does it have to do with the fact that I am digitizing using a Digital8 camera and Firewire?
    Video 8 is analog, there are no pixels. Analog NTSC video has 525 total scanlines, of which 486 lines (give or take a couple of lines) are used for so-called "active area". This number is codified in D1 and Rec. 601 standards. The frame size follows a standard defined forty years ago for representing digital video while taking into account the ease of conversion between 625/50 and 525/60 broadcast TV systems that existed back then. 480 is a slight modification that has become common thanks to DV and DVD. 480 is divisible by 16, which is useful when using 16x16 macroblocks (720 is also divisible by 16).

    720x480 video represents a 4:3 frame, but there is a kink - black borders on the sides are not included in the 4:3 frame. Dare I suggest my own short video on the topic? (TL;DR: crop 16 pixels on the sides and consider the remaining 704x480 frame to have 4:3 proportions).
    Quote Quote  
  18. Capturing Memories dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Member Since 2005, Re-joined in 2016
    Search PM
    It's not that simple, the active video area is not always in the center so cropping 8 pixels on each side blindly is not going to look good, Also it is not always 704, sometimes more, sometimes less, It's usually anywhere from 694 to 712 depends on the format and the recorder/camera that recorded the content. That's why it is always to start with a capture card that doesn't have any cropping or resizing capabilities on the fly, then manually do the cropping after the capture.

    I found this to change from one TV commercials to another even on the same tape, That's why I average the cropping area leaving some minor black borders on some commercials, it's time consuming to do them individually.
    Quote Quote  
  19. Member Cornucopia's Avatar
    Join Date
    Oct 2001
    Location
    Deep in the Heart of Texas
    Search PM
    @just4747,
    You are mentioning terms "lossy" and "lossless" in ways that make me think you don't quite understand them.

    Here's the breakdown:

    When you digitize, the action/process is alway lossy, whether a little or a lot, unless you were to digitize using a huge highbit HDR floating point format. Normal formats have some form of quantization, which (rounding, truncation, etc) is a form of loss because the inexact analog level needs to be assigned an exact digital level, and there is almost always a small amount of error there.
    This normally ONLY occurs at the actual point of digitization.

    When you play out digital stuff, even to analog material, it is not normally a lossy process, as those exact digital levels have no trouble with exact digital levels, nor do they have trouble with "inexact" analog levels, as analog is a continuum so any of those levels are good.

    On top of those basic processes, you may have to convert formats. For example YUV vs. RGB color spaces, converting between the various color subsampling formats (4:1:1, 4:2:0, 4:2:2, etc), and when resizing and changing framerates or interlacing. The color space conversions can occasionally be lossless, but are usually lossy. The color subsampling conversion is rarely lossless (only special cases), and the resizing and framerate/interlace conversions are NEVER lossless (not counting TC/IVTC here).

    While that stuff is in the analog realm, if you transfer/copy, store, distribute, etc, you will incur loss.
    While that stuff is in the digital realm, within the bounds of standard error correction, you will NOT incur any loss with transfer/copy, storage, distribution, etc.

    And then you have not just processes that can be lossy or lossless, you have encoding formats that can be lossy or lossless.
    Uncompressed is lossless. "Lossless compressed" is lossless. These are giganticly huge, and huge, respectively.
    Everything else is lossy. These can be huge, large, medium, small, or tiny depending on the choice of codec and bitrate. Depending on those choices, they also determine the amount of loss, which isn't always linear.
    "Virtually lossless" lossy codecs are large, but the amount of loss is small enough to either not be noticeable, or be barely noticeable, even with multiple generations of re-encoding/conversion. Other lossy codecs (which include DV, MJPEG, and most MPEG family) are medium or small size, but their loss is noticeable, and cumulatively noticeable if redone in multiple generations.

    So to sum up, with an example related to what you've been talking about:
    You have an Analog tape and you digitize it with a USB converter to a "lossless" format. The digitization that occurs in the USB device (or in the software at the PC, facilitated by the USB device) incurs a small amount of loss due to quantization and usually due to color subsampling. If no other conversions occur (colorspace, resizing, framerate/interlacing), then you will have no other losses. But if you started with a 525/29.97i scanline analog signal, you would end up with a 486/29.97i (yes, interlaced) or 480/29.97i digital signal. If you needed to deinterlace, or change the rate or resize, crop*, etc, you WILL be incurring loss of quality, even if your result has a "resolution" or framerate designation higher than your original (because you cannot create new, meaningful detail out of thin air - "AI" be damned).

    If you took that analog tape and put it instead through a DV camera or DV converter box, it is digitizing it within that device (and losing quality due to the same quanitzation and color subsampling effects that the other process had), but it also is encoding it using a lossy type of codec ("DV"). Whether it is stored on tape or not, that doesn't matter, because remember those digital transfer/copy/storage processes are lossless within the bounds of error correction. When playing back (or playing through), this already-digital signal is still just being transferred. So no loss in that part of the process (but loss overall because it already lost something during the digitization, and more during the lossy encoding). As long as it doesn't get converted any further, it will not incur any further loss. But it will be a 480/29.97i signal (DV always saves 525 as 480 lines). And yes, that means it will still be interlaced as well.

    If you edit the first example or the 2nd, both can still not incur further loss IF you do not change the internal values of the pixels or do a conversion. So simple cut/assemblage edits only rearrange the stream of frames, and don't convert or re-encode, so no loss. However, color correction or subtitling (or other types of mixing/overly, or cropping as some suggest) DOES change the values of the pixels, and will require re-encoding, even if nothing else is done to it. Thus, more loss (another generation of DV compression) if you used the 2nd method, but NO more loss, if you used the 1st method (and continued to maintain it).

    Note: IIRC, you have mentioned deinterlacing in some of those posts, and that is a LOSSY conversion (ALWAYS). The best versions of algorithms are quite good, but none are without loss. Personally, depending on the type of (edit/distribution) workflow, I recommend that you either do a deinterlace at near the very beginning or near/at the very end, so that this type of loss is only incurred once. But you have to be careful that you understand the state of your files so they can be properly edited/processed, etc.

    If you STARTED with a DV camera to begin with, it is already digitized and DV-encoded (loss, loss). So you would NOT want to play out via analog (NO LOSS) and then back in to another form of digitization (LOSS, and possibly more LOSS), because you would be unnecessarily compounding loss.

    That's a lot to take in, and hopefully it cleared things up rather than muddy things up.

    Scott

    *certain kinds of crop can be lossless, but they rarely are.
    Last edited by Cornucopia; 9th May 2022 at 16:16.
    Quote Quote  
  20. Member Skiller's Avatar
    Join Date
    Oct 2013
    Location
    Germany
    Search PM
    Originally Posted by dellsam34 View Post
    It's not that simple, the active video area is not always in the center so cropping 8 pixels on each side blindly is not going to look good, Also it is not always 704, sometimes more, sometimes less, It's usually anywhere from 694 to 712 depends on the format and the recorder/camera that recorded the content.
    You cannot judge if a video is centered by looking at the blanking at the sides. A video produced in the analog times can be perfectly centered (meaning whatever was in the very center during the shoot or post is still exactly there) but there may be a huge border on the left and barely any on the right. Without a test pattern or some hint you cannot be sure. Unless it is obviously shifted I would always blindly crop to the center. That's how it would be displayed on a TV as well. Think of it this way: what was once actual video might have simply been overridden with black at some point (quite typical, happended all the time). This doesn't shift the position or change the size.


    Originally Posted by dellsam34 View Post
    I found this to change from one TV commercials to another even on the same tape, That's why I average the cropping area leaving some minor black borders on some commercials, it's time consuming to do them individually.
    In my opinion you should just do what was indended to happen: overscan. Crop to emulate overscan. I would probably crop to the center of 660x450 for a rather small overscan emulation but it will eliminate all blanking areas and look more like it would have on a TV.
    Quote Quote  
  21. Capturing Memories dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Member Since 2005, Re-joined in 2016
    Search PM
    That's in theory but in practice it's rarely in the center, I'm talking from experience dealing with several analog video formats not from a text book.
    Quote Quote  
  22. Member pchan's Avatar
    Join Date
    Mar 2003
    Location
    Singapore
    Search Comp PM
    In the past where data storage was expensive, we focused on compression with highest video quality and mode of delivery i.e. VCD, DVD and HD. Now, H.264/H.265 becomes the norm, and by looking at the DVD/Bluray players and media in the market, it's almost dead. Cheap portable USB hard disks are in terabytes so keep backup in the highest quality is no longer an issue. A 2TB USB hard disk can store about 150 hours of video in DV format. Sharing video is easy e.g. FB, Youtube, Whatsapp etc and mp4 is supported everywhere. No longer have worry about those RMVB monkey craps. Cheap set top boxes can playback HD format.

    So, it's trivial, backup in the highest quality e.g. DV, AVCHD format. DV to mp4 conversion no long takes hours. It's in minutes if not seconds with current CPUs. HandBrake(mp4 converter) is a standard feature in CPU benchmark.
    Quote Quote  
  23. Capturing Memories dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Member Since 2005, Re-joined in 2016
    Search PM
    DV and AVCHD are not the highest quality and DV to h.264 conversion is destructive, If storage is not an issue anymore as you stated capture to lossless AVI and encode to h.264.
    Quote Quote  
  24. Formerly 'vaporeon800' Brad's Avatar
    Join Date
    Apr 2001
    Location
    Vancouver, Canada
    Search PM
    Originally Posted by just4747 View Post
    was mainly wondering why so many people suggest using the same method I am now (DV via a camera) but with VDub/HuffyUV vs WinDV like I am currently.
    Originally Posted by just4747 View Post
    I was referring more to what I am doing/using now vs using that same setup but with the often recommended VirtualDub/HuffYUV. I keep reading people recommending to use those for better/lossless capture with the same DV/Firewire setup I am using now. Is it worth it or should I just stick with WinDV? I've read countless times that doing DV transfer the way I am now via just WinDV (lossy) is not as good vs using VDub/HuffyUV which will be lossless.
    Must be a communication breakdown somewhere. Hopefully you were able to digest Cornucopia's post. If you want to go the "analog → Digital8/DV passthrough" route, transferring the bits via WinDV is as good as that method gets.

    Complicating this simple passthrough conversion setup by attempting to add in a "lossless/Huffyuv" step would in fact incur loss, ironically, on top of the initial DV compression loss. (Because Huffyuv doesn't support 4:1:1, but this is getting way into the weeds.)


    And one more question if you don't mind - I mentioned that the files I am getting via WinDV now are still saying 720x480 for resolution but 736x480 for buffer dimensions. Is this normal or is this maybe not accurate because VLC is guessing?
    I suspect the "buffer dimension" is only of use to programmers and can be safely ignored.
    My YouTube channel with little clips: vhs-decode, comparing TBC, etc.
    Quote Quote  
  25. Originally Posted by just4747 View Post
    ......If all I want to do with the videos after DV capturing is to make slight edits (cutting, trimming, joining or removing scenes, titles, color correction, etc.. in Premier Pro, .....
    Sidenote: Apparently the OP plans to do some "slight" editing in Premier Pro. I am not familiar with PP but with respect to the lossless/lossy discussion the editing may involve a colorspace conversion (something like limited range (?) DV YUV capture -> full range RGB for editing -> limited range YUV for final export) and possibly re-compress the video or at least part of it. Means possibly another source for introducing "losses". Someone familiar with PP may have more on this.

    Edit: I just noticed that Premier Pro has a YUV capable timeline so at least the potentially lossy YUV->RGB conversion issue should not apply when importing the DV capture in PP.
    Last edited by Sharc; 10th May 2022 at 04:22.
    Quote Quote  
  26. Captures & Restoration lollo's Avatar
    Join Date
    Jul 2018
    Location
    Italy
    Search Comp PM
    Unless it is obviously shifted I would always blindly crop to the center.
    What I normally do to center a 720x576 capture is to crop all back borders, then add same amount on the left and the right side and on the top and bottom side to build the 704x576 frame. Example:

    Code:
    # cropping 
    	crop_left=8	# | removal of black borders at the right, top and left and of head switching noise at the bottom
    	crop_top=2	# | 720-(8+16)x576-(2+10)=696x564
    	crop_right=16
    	crop_bottom=10
    video_crop_addborders=video\
    .crop(crop_left,crop_top,-crop_right,-crop_bottom)\
    .addborders((crop_left+crop_right)/2-8,(crop_top+crop_bottom)/2,(crop_left+crop_right)/2-8,(crop_top+crop_bottom)/2)
    Quote Quote  
  27. Formerly 'vaporeon800' Brad's Avatar
    Join Date
    Apr 2001
    Location
    Vancouver, Canada
    Search PM
    [Off-topic discussion. I'm indenting so the OP can more easily ignore my OT post...]
    Originally Posted by Skiller View Post
    Originally Posted by dellsam34 View Post
    It's not that simple, the active video area is not always in the center so cropping 8 pixels on each side blindly is not going to look good, Also it is not always 704, sometimes more, sometimes less, It's usually anywhere from 694 to 712 depends on the format and the recorder/camera that recorded the content.
    You cannot judge if a video is centered by looking at the blanking at the sides. A video produced in the analog times can be perfectly centered (meaning whatever was in the very center during the shoot or post is still exactly there) but there may be a huge border on the left and barely any on the right. Without a test pattern or some hint you cannot be sure. Unless it is obviously shifted I would always blindly crop to the center. That's how it would be displayed on a TV as well. Think of it this way: what was once actual video might have simply been overridden with black at some point (quite typical, happended all the time). This doesn't shift the position or change the size.
    Originally Posted by dellsam34 View Post
    That's in theory but in practice it's rarely in the center, I'm talking from experience dealing with several analog video formats not from a text book.
    Originally Posted by lollo View Post
    Unless it is obviously shifted I would always blindly crop to the center.
    What I normally do to center a 720x576 capture is to crop all back borders, then add same amount on the left and the right side and on the top and bottom side to build the 704x576 frame.
    Skiller's point, as I understand it, was that this is technically a distortion of the original signal. An accurate presentation of the video as seen on some theoretical, perfectly-calibrated CRT of the time would have the horizontal center precisely in the center of the horizontal sync pulses, no? And once digitized, the picture center is defined by spec. He's speaking of the center as originally "framed up", not in terms of where the edges of blanking sit in a final broadcast after it's passed through many signal chains.
    (Charles Poynton, Digital Video and HD: Algorithms and Interfaces, 1st Ed.)
    Of course, if you're going as far as shifting the vertical center of a 576 capture after head-switch noise crop, you don't care about this little inaccuracy.
    (None of this really matters to me either, personally. I'm happy to accept both some shift and a small percentage of aspect ratio distortion. Even the best BVMs/PVMs of the time weren't perfect and couldn't manage the precision of our flat panels that base their geometry on subpixels...)
    Last edited by Brad; 10th May 2022 at 07:36.
    My YouTube channel with little clips: vhs-decode, comparing TBC, etc.
    Quote Quote  
  28. Captures & Restoration lollo's Avatar
    Join Date
    Jul 2018
    Location
    Italy
    Search Comp PM
    Of course, if you're going as far as shifting the vertical center of a 576 capture after head-switch noise crop, you don't care about this little inaccuracy.
    Good point (as always from you), but yes. I prefere a shift compared to what the picture looks like in a CRT that I do not have anymore, than having unbalanced black borders on modern screens
    Quote Quote  
  29. Member Skiller's Avatar
    Join Date
    Oct 2013
    Location
    Germany
    Search PM
    Originally Posted by dellsam34 View Post
    That's in theory but in practice it's rarely in the center, I'm talking from experience dealing with several analog video formats not from a text book.
    I'm not talking theoretical either. I saw exactly what you are talking about (borders change with every commercial for example). Again, you cannot tell if it is centered by looking at the size of the blanking left and right. It doesn't mean anything. In most cases one simply cannot do anything about it due to a lack of reference – but frankly, does it matter?


    Originally Posted by Brad View Post
    He's speaking of the center as originally "framed up", not in terms of where the edges of blanking sit in a final broadcast after it's passed through many signal chains.
    That's exactly what I'm talking about, thanks Brad.



    Sorry for hijacking the thread guys.
    Last edited by Skiller; 10th May 2022 at 08:10.
    Quote Quote  
  30. Captures & Restoration lollo's Avatar
    Join Date
    Jul 2018
    Location
    Italy
    Search Comp PM
    Sorry for hijacking the thread guys.
    No problems on my side, your posts are always interesting!
    Quote Quote  



Similar Threads

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