Comparison of S-Video and DV capture of Video8 - which of these preferred?

[just4747]

Hi all - long time lurker.

I have been reading and testing a million different capture types, settings, programs, etc to get the best digital capture I can of approx. 40 family Video8 tapes with the available hardware.

Currently I am comparing two captures of a clip and trying to figure out which method I should be working with/which looks better to most people. Details of hardware and software:

-Sony Handycam Digital8 DCR-TRV310 Camera and videos are on regular analog Video8 tapes
-(Method 1) S-Video from camera to cheap USB 2.0 capture device with S-Video/RCA inputs
-(Method 2) DV-Out from camera to PC Firewire card with Firewire cable
-VirtualDub 1.10.4 on Windows 11 PC

See screen shots below - I figured just screenshots alone will tell a knowledgeable person which one is on the right track, if either.

I have captured a short clip using each above method to compare, both captured using default codecs in VirtualDub and 720x480 format (though the DV one came out with 736x480 buffer somehow, so maybe ignore the stretch it has). To me, the S-Video clip looks slightly sharper but noisier and the DV/Firewire one looks a little smoother/less detailed with less noise. The DV clip size is 200MB vs the S-Video at 90MB.

Am I on the right track with either of these and the way I am trying to capture? I am going nuts with all the things I've read and suggestions, it's just overwhelming. First I had tried using VDub with the suggested HuffYUV codec and YUY2 format, but VDub gives me an error when I try to use either of those for capturing using either output type, which is why I went back to using their default codecs for each.

Any feedback and suggestions appreciated!


EDIT: Actual clips embedded for DL in this post:
https://forum.videohelp.com/threads/405649-Comparison-of-S-Video-and-DV-capture-of-Vid...ed#post2655899


Screenshots:

[Attachment 64656 - Click to enlarge]

[Brad]

The S-Video capture is trash IMO due to over-sharpening. The noise doesn't bother me, but the halos do.

With the USB capture device, in VirtualDub capture mode try Video -> Levels. Neutral Sharpness level is often 0. Sometimes it's the middle setting.

The Firewire version looks remarkably natural to me. I'm surprised it's Video8.

[just4747]

The S-Video capture is trash IMO due to over-sharpening. The noise doesn't bother me, but the halos do.

With the USB capture device, in VirtualDub capture mode try Video -> Levels. Neutral Sharpness level is often 0. Sometimes it's the middle setting.

The Firewire version looks remarkably natural to me. I'm surprised it's Video8.

Interesting, so is the noise coming from the sharpening that the S-Video has? What is the sharpening caused by if I'm just capturing it as-is? Didn't adjust or change any capture settings.
And what are the halos you refer to? I don't see anything obvious.

Also I tried to adjust levels in VDub and in OBS and both have the sliders grayed out/unavailable when using the USB capture device and this camera.

Doesn't everyone say DV is lossy and that S-Video is preferred for a situation like this though (capturing analog video to digital)? I know that results are what matter in the end but I've read that several times now, though some say to use DV like you are.

[ConsumerDV]

I did not know one could use VirtualDub for capturing DV. Does it preserve the original bitstream? Judging by non-standard frame size, it seems it does not. Do you have anything against WinDV or Scenalyzer?

Contrary to many members of this forum I prefer digitizing Video8 and Hi8 through a Digital8 camcorder: you get reliable quality in an industry-standard codec with correct metadata without fuss.

Looking at your S-Video capture, I suppose you are going to fix the incorrect aspect ratio.

[Skiller]

The DV clip size is 200MB vs the S-Video at 90MB.

That's odd, the DV capture should be smaller, not the other way around. What codec did you use in VirtualDub?

The DV capture looks like it was shot with a native Digital8 or MiniDV camcorder to me – it's remarkably good if this really is Video8 on the tape.

The S-Video capture has some major issues that Brad named.

[dellsam34]

See screen shots below - I figured just screenshots alone will tell a knowledgeable person which one is on the right track, if either.

Not that screenshot, A screenshot of a pause from the screen doesn't give any info at all, A screenshot from the player saved as snapshot using no processing could be helpful though a short sample from the native files is the best way to tell the full story, Some temporal artifacts especially on DV are not shown in a freeze frame.

[just4747]

I did not know one could use VirtualDub for capturing DV. Does it preserve the original bitstream? Judging by non-standard frame size, it seems it does not. Do you have anything against WinDV or Scenalyzer?

Contrary to many members of this forum I prefer digitizing Video8 and Hi8 through a Digital8 camcorder: you get reliable quality in an industry-standard codec with correct metadata without fuss.

Looking at your S-Video capture, I suppose you are going to fix the incorrect aspect ratio.

All I was reading about was using VDub for DV transfer - I saw mention of that over any other program honestly. I just hear that it's a really good way to get high quality DV captures and lossless ones if needed with certain codecs and such. I've really never used WinDV or Scenalyzer so not sure how to compare them all or if I should be using one over the other.

Also, I am using a Digital8 camcorder for transferring these Video8 tapes, so same as you said - just trying to figure out DV vs. S-Video and the right software, etc.. Still many many options and methods regardless of how. Do you have a recommendation of exactly how to do it using one of these outputs from the camera and which software/method?

And what did you mean by non-standard frame size? The screen shots are just the clips playing in VLC player at original size and the DV one is slightly different because it exported with a 736x480 canvas for some reason even though the video is 720x480.

[ConsumerDV]

I am not going to question using VDub for DV capture, I just did not know it could do that. As long as it gets the bitstream from a Firewire port and saves it as a proper DV-AVI file, it is all good. It should come as 720x480, you just mentioned 736x480, which threw me off.

[just4747]

The DV clip size is 200MB vs the S-Video at 90MB.

That's odd, the DV capture should be smaller, not the other way around. What codec did you use in VirtualDub?

The DV capture looks like it was shot with a native Digital8 or MiniDV camcorder to me – it's remarkably good if this really is Video8 on the tape.

The S-Video capture has some major issues that Brad named.

Yeah not sure why it would be such a huge size difference but again I just used the default codecs and let it record - both the same 1 minute clips from the same tape.

And I swear it's Video8 tape going through a Digital8 camera. Maybe it looks so good just bc it's a screen shot? I forget what it called the default compression for DV capture (I don't have it hooked up right now but it was whatever it defaults to/the only option), but the default compression setting used for the S-Video in VDub says "No recompression: MJPG" and has "FOURCC code" and "Drive name: uncompressed" next to it.

Here are both the tape and my camera FYI:



[Attachment 64661 - Click to enlarge]

[just4747]

I am not going to question using VDub for DV capture, I just did not know it could do that. As long as it gets the bitstream from a Firewire port and saves it as a proper DV-AVI file, it is all good. It should come as 720x480, you just mentioned 736x480, which threw me off.

Ah ok - yeah not sure why either - I'll have to check the settings and try again with the exported size.

[just4747]

See screen shots below - I figured just screenshots alone will tell a knowledgeable person which one is on the right track, if either.

Not that screenshot, A screenshot of a pause from the screen doesn't give any info at all, A screenshot from the player saved as snapshot using no processing could be helpful though a short sample from the native files is the best way to tell the full story, Some temporal artifacts especially on DV are not shown in a freeze frame.

I just thought the two looks were so wildly different that to people who are versed in this stuff, there would be an obvious one of them at least going in the right direction (sounds like DV is it). I can def provide a short video sample though of each if you can help me critique that way.

Both are AVI files - how can I re-copy like a 15 second part from each clip and drop the audio while keeping the same quality and look/detail each have now? Is that possible?

Thanks!

[Brad]

Both are AVI files - how can I re-copy like a 15 second part from each clip and drop the audio while keeping the same quality and look/detail each have now?

lollo2 linked to this old guide recently; looks pretty good. Just modify it by choosing No audio.

Then Upload files/Manage attachments on the forum (lower left of reply box).

[lollo]

Am I on the right track with either of these and the way I am trying to capture?

To make a right comparison you should use a good USB capture device, and, as Brad said, disable all sharpening in player, capture device and capture software. Capture 8bit YUY2 4:2:2 lossless.

I ignore if VirtualDub saves the DV stream coming from the camera firewire port or if it is decoding/encoding to DV (I never used it for this purpose). Just use WinDV or Scenalyzer.

[Sharc]

And what are the halos you refer to? I don't see anything obvious.

The high-contrast dark/bright transitions at sharp edges, like at the back of the boy's shirt.

The MJPG compressed S-video capture has an excessively (edge-)sharpened U plane.

[Skiller]

At this point there are some oddities and questions remaining about your capture processes.

Assuming both are about the same length, an S-Video capture should be much larger in file size than an actual DV capture; even with M-JPEG codec it should at least not be vastly smaller. I'm thinking you may not be capturing the actual DV bitstream. VirtualDub never used to be suited for DV capture so that's a strong indication for this theory. For DV capturing you should use WinDV or Scenalyzer.

If you cannot get HuffYUV to capture it might be because your driver is set to an unsupported color space. In VirtualDub in capture mode check what it says if you click "Video" then "Capture pin...".
Color space needs to be "YUY2".
Do not capture with any weird "default" codecs.

[just4747]

And what are the halos you refer to? I don't see anything obvious.

The high-contrast dark/bright transitions at sharp edges, like at the back of the boy's shirt.

Ah got it thanks. Yeah I guess it's easy to see that the DV is much more natural looking. I guess it's easy for someone that doesn't have a trained eye to think the sharpened one looks better when it clearly does not in this case.

[just4747]

Both are AVI files - how can I re-copy like a 15 second part from each clip and drop the audio while keeping the same quality and look/detail each have now?

lollo2 linked to this old guide recently; looks pretty good. Just modify it by choosing No audio.

Then Upload files/Manage attachments on the forum (lower left of reply box).

That worked great - thanks. The clips I just made have the same quality as the full clips.

I attached the two clips below, cut down to 16 seconds each with no audio. I still have to address some other comments here how I did the captures but these are the clips from the OP.

Thanks all!

[just4747]

Am I on the right track with either of these and the way I am trying to capture?

To make a right comparison you should use a good USB capture device, and, as Brad said, disable all sharpening in player, capture device and capture software. Capture 8bit YUY2 4:2:2 lossless.

I ignore if VirtualDub saves the DV stream coming from the camera firewire port or if it is decoding/encoding to DV (I never used it for this purpose). Just use WinDV or Scenalyzer.

Thanks, but since most people are making me lean towards the DV capture, I guess I won't need to use a USB capture device for S-Video anyway and will just be using the FireWire PCI card, so I guess that simplifies it a bit.

I will definitely be trying WinDV or Scenalyzer to capture DV though, hopefully it's simpler and less issues. Is either recommended over the other in my kind of situation?

[Skiller]

I will definitely be trying WinDV or Scenalyzer to capture DV though, hopefully it's simpler and less issues. Is either recommended over the other in my kind of situation?

It is simpler. DV capture means you are just copying the bits transferred via Firewire to your hard disk. I would go for Scenalyzer. The interface is not the best but it has all you will ever need.


That DV sample is actually a proper DV bitstream. It looks fine, in my opinion you should go that route. The S-Video capture has been deinterlaced, is over-sharpened and has lots of noise (wondering where that comes from).

[lollo]

...so I guess that simplifies it a bit.

Yes, but you will loose quality compared to first scenario (good USB capture device, 8bit YUY2 4:2:2 lossless capture), in particular if you plan restoration.

Is either recommended over the other in my kind of situation?

They are equivalent in term of "quality" of the transfer. Scenalyzer has more features, i.e. split the capture at scene change, if you like that.

edit: Skiller arrived first

[just4747]

...so I guess that simplifies it a bit.

Yes, but you will loose quality compared to first scenario, in particular if you plan restoration.

Is either recommended over the other in my kind of situation?

They are equivalent in term of "quality" of the transfer. Scenalyzer has more features, i.e. split the capture at scene change, if you like that.

edit: skiller arrived first

When you say lose quality compared to the first scenario - do you mean S-Video output? Sorry just want to confirm what you are referring to because everyone is telling me to def go the DV transfer route.

[just4747]

I will definitely be trying WinDV or Scenalyzer to capture DV though, hopefully it's simpler and less issues. Is either recommended over the other in my kind of situation?

It is simpler. DV capture means you are just copying the bits transferred via Firewire to your hard disk. I would go for Scenalyzer. The interface is not the best but it has all you will ever need.


That DV sample is actually a proper DV bitstream. It looks fine, in my opinion you should go that route. The S-Video capture has been deinterlaced, is over-sharpened and has lots of noise (wondering where that comes from).

I will definitely try this then. 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.
Any idea though why my DV capture is so much larger than the S-Video one? It sounds like I did it wrong and maybe when I go to one of the other programs it will solve it, but still curious.

Thanks.

[Skiller]

Yes, but you will loose quality compared to first scenario, in particular if you plan restoration.

Right, that's due to DV compression. But for a lossless S-Video capture to have an advantage over the DV one all other aspects would first have to meet the quality the Digital8 camcorder's transfer. And looking at that S-Video sample this seems very far away. Probably needs other hardware.

Any idea though why my DV capture is so much larger than the S-Video one?

It's the M-JPEG codec (should not be used). Bitrate in that sample is only about 12.5 MBit/s, which is about half of that of DV. In other words, it is compressed like mad (but that's not the main reason it looks worse).

[lollo]

The hardware (camera player) is the same in both scenarios if I am not wrong, so lossless S-Video is preferable anyhow. But you are right about the sample provided, it has some bad in it, I ignore why.

[lollo]

When you say lose quality compared to the first scenario - do you mean S-Video output?

Yes, but with one of the recommended card!

[Skiller]

The hardware (camera player) is the same in both scenarios if I am not wrong

For S-Video the Digital8 camcorder is merely the VCR playing the tape. just4747 says he uses a "cheap USB 2.0 capture device with S-Video/RCA inputs" and I would say this is why it looks so bad.

[lollo]

and I would say this is why it looks so bad.

Most likely, that's why I suggested a "recommended" card instead. But DV route can be acceptable for his goals as well, as you properly said.

[ConsumerDV]

I will definitely be trying WinDV or Scenalyzer to capture DV though, hopefully it's simpler and less issues. Is either recommended over the other in my kind of situation?

Whichever works for you. The output is the same, it is whatever is sent over FireWire. Lately, WinDV would not recognize that my camcorder is connected, Scenalyzer seems more robust. In terms of UI and features, I don't need all these things that Scenalyzer offers, to me WinDV is enough.

Yes, but you will loose quality compared to first scenario (good USB capture device, 8bit YUY2 4:2:2 lossless capture), in particular if you plan restoration.

Capturing 4:2:2 into high-quality card through outboard TBC, yeah, yeah, yeah. But for someone who simply wants to get their home videos off a tape, DV route is straighforward, predictable, reliable and provides known quality without fiddling with various settings.

When you say lose quality compared to the first scenario - do you mean S-Video output? Sorry just want to confirm what you are referring to because everyone is telling me to def go the DV transfer route.

Here is one of the recent discussions... The OP has left the thread by the end of the first page Mind you, I still think that DV->FireWire route is better for, shall I say, normal people.

Any idea though why my DV capture is so much larger than the S-Video one? It sounds like I did it wrong and maybe when I go to one of the other programs it will solve it, but still curious.

It is not that DV capture is larger, it is your SVideo capture is smaller 12.5 Mbit/s is way too low for MJPEG (why MJPEG?). Also, as others have mentioned, it is deinterlaced into 30p, so you have lost half of the images, meaning you lost half of the spatial resolution and half of the temporal resolution. Also, why 30.00 fps instead of 29.97 fps? See, in the case of DV route you do not need to think about it.

[nicholasserra]

+1 to all of that ^

I capture 4:2:2 lossless via TBC , capture card etc etc, but it's not for everyone and can be a real pain until you have all the kinks worked out. Firewire > DV is the easy mode and it's "good enough" for most people, and won't have the same issues as those junk $15 amazon usb card that capture to h264.

[energizerfellow]

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.

If you have a Thunderbolt port handy, try the Apple Thunderbolt 3 (USB-C) to Thunderbolt 2 Adapter in combination with the Apple Thunderbolt to FireWire Adapter, which is both reliable and works in Windows.