Best s-video to USB capture transfer device?

[WarbirdVideos]

I'm capturing a number of VHS tapes of with historical video on them recorded in the the early '90s on standard VHS. I'm using an S-VHS JVC HR s4600u deck, outputting the s-video connection to a Pinnacle 710-usb interface box, Windows 7, capturing with VirtualDub using the Magic M8Y0 codec. The results are OK, but not the best.

So I routed the s-video output of the S-VHS deck into a Sony DSR-30 DVCAM deck and recorded it to DVCAM, then transferred the DVCAM to the computer via the firewire port. The results were better than going through the Pinnacle 710 interface card.

So the question is: What is the best s-video to USB capture device to use from the playback deck to the computer, rather than using a DVCAM sub-master?
Thanks!
Steve

[lordsmurf]

Why were you using a non-standard codec, not Huffyuv?

I need to visually see what "better" means to you. Because DV will be very lossy. Therefore this makes no sense.

You're also lacking any TBCs, neither line nor frame. That is a problem. So results from either method will be extremely inferior, with all sorts of visual flaws and signal issues.

Again, samples needed.

[Hushpower]

I think you should be using 4:2:2 for captures:

MagicYUV Codec options.jpg

Also, have a look at this:

https://www.digitalfaq.com/forum/vid...html#post79495

LS, I use MagicYUV because my NLE currently can't handle the old codecs such as HUFF and LAGS.

[lordsmurf]

MagicYUV usually does not handle capturing very well. Too much overhead for casual systems. (I use it for post-capture intermediary, mostly Mac<>Win projects.)

But yes, 4:2:2 for sure needs to be used.

I still want to see visually what "better" means. In video terms, it's really a nothing description. We need to put some descriptors on the observations.

[WarbirdVideos]

Thanks for getting back to me!

Attached are still photos of the Pinnacle 710 and the DVCAM capture through VituralDub.

- I have an old TBC, but there was only quality loss using it. I don't see any timing errors though. Would there be a benefit in using a TBC if the video plays back solid and steady?

- I used the Magic codec because it's recommended by HappyOtterScripting and it seems a bit smaller. I've used Lagarith and Huffyuv in the past. I'm not sure how much quality gain there is by using 25 gigs per hour for VHS. HD video uses about 17 gigs per hour. Is there data on this? There's so many fewer pixels in VHS

I'm thinking I may knock the file size down with handbrake, but not sure. I might just produce the segments and upload them as an .mp4. The final vids will be deinterlaced, converted to 60p, Y/C delay corrected, and upscaled to a larger size - perhaps double or a bit smaller.

[latreche34]

Stills don't tell the whole story but from them I don't see why you think DV is better, Yes the contrast is a little higher but that comes at cost of crashing blacks.

[WarbirdVideos]

Hushpower: I used 4.2.2 on my Hi8 transfers, but honestly, I don't know if there's a benefit to using 4.2.2 over 4.2.0 on low quality VHS, or single chip Hi8 for that matter. It just creates larger files. As I recall, 4.2.2 will create files of 32 gigs per hour or and 4.2.0 is about 22gph. I can't tell the difference. In fact, I've run them through handbrake and the file size drops to 8 gigs per hour and it looks the same.

[Hushpower]

@WarbirdVideos, the idea is that you capture at the absolute best quality and then process/restore, then encode/export down to MP4 for final use/viewing, which will be 4:2:0. That's why 4:2:2 is recommended for capture.

Regarding Handbrake exports, you should be able to get down to around 4gb per hour (H264) with no visible quality loss. Use a CRF of 16 or so.

[lordsmurf]

We need to see small clips, not just stills. However, there was some good info to start with.

4:2:0 is a compression comparable to DVD-Video. It still tosses out color data, but not as damaging as 4:1:1 from that DVCAM.

I've never liked it when somebody says "low quality VHS". The reason for the low quality is rarely due to the VHS format, or the VHS tape, but instead due to the cheap/bad equipment in use (especially crummy consumer VCRs). In this case, this is definitely true. The chroma noise is obnoxious.

For example, those blinds are obviously white (ignore the whole-frame blue tint). But in the video, you see flecks of red, blue, green, yellow, etc. That's chroma noise, color that is not actually there is inserted into the video. Quality VCRs with line TBC would run a pass filter that removes the noise. You'd have white blinds again, not blinds where a crayon box barfed on it.

Noise hides artifacts, because quality already sucks. Sucking more doesn't affect it as badly. It's like a bird pooping on a pile of cow dung.

But here's the problem: That chroma noise can (sort of) be filtered post-capture. Some issues can only be corrected in hardware, but chroma noise isn't one of those. So now that you remove the chroma noise -- uh oh! All the other damage is now obvious, because the video was made worse. Both under the chroma noise, and even due to the presence of the chroma noise (compression hates noise!).

The DV here definitely lost color, in addition to blowing out highlights. Look closely at the left-most and right-most flags in the image. And while you can give an excuse that it's "not a big deal", remember that this is just a still. Things get vastly worse in motion. The "Golden Gate" yellow also has some infamous DV tint changes going on, ot's both blurred and cooked. Lots of shadow ringing (halo, outline) is added in/around the picture frames.

The DV is not better whatsoever here. It's lossy.

And while some people may give excuses to dismiss the quality loss ("good enough", etc), remember that this is just a tiny preview. Viewed large, even half-blind grandmas noticed the quality loss. You lost detail, color changed. "Why do we all look sun burned?"

I hate the cop-out attitude of "oh well, just VHS". Because then it really wasn't worth converting at all. I'm not seeing that here, but just wanted to mention it, for any future readers.

As far as SD vs. HD file size, it's not 1:1. Lossless HD is huge, while compressed HD is not. Don't compared compressed HD to lossless SD. Not a valid comparison.

[WarbirdVideos]

Has anyone run across a res chart with colors over the years to show the real differences of using different methods of capture with different codecs? I attached .bmp images rather than jpg. It is very close.

[lordsmurf]

That would be invalid, and here's why:

The palette for DV is fixed at 720x480. But that means nothing in terms of the true image resolve, which was determined by the recording camera eons ago. It's true of both DV tapes and VHS tapes. DV is at least digital tape, while the VHS lost a bit more on the media write.

There are many "pixel peeps" on DV vs. lossless, going back at least 20 years. Both here and at a VH.

It's not close. It's just not. We're talking measurable double-digit % losses in the chroma.

If you have double-digit % losses in the stock market, you'd not claim it was close (or "fine" or whatever), you'd be freaking out (in either way). Sort of like NVDA this past week. Was the new price close to the old price? Only 30%, right? Math matters. It can be artistic at times, but it still matters.

[Hushpower]

Wa La! Coupla clicks... Neat Video.

Porter DVCAM Neat.jpg

I'm not sure those blinds are white... Clearly the book is white, and those blinds are not the same colour as the book.

[traal]

Quote:

Originally Posted by WarbirdVideos

Would there be a benefit in using a TBC if the video plays back solid and steady?

With a good TBC? Possibly. See this post for a visual comparison.

Quote:

Originally Posted by lordsmurf

For example, those blinds are obviously white (ignore the whole-frame blue tint). But in the video, you see flecks of red, blue, green, yellow, etc. That's chroma noise, color that is not actually there is inserted into the video.

Or it could be caused by a single-CCD (or older vidicon type) camcorder, low light levels, and fluorescent lighting. But we'll never know without a good VCR and a good capture card.

[WarbirdVideos]

@lordsmurf: So what TBC and playback deck do you recommend?

I have 40 years of professional shooting, lighting, sound, editing, and engineering experience, so I'm not a newbie with home movies. But this is what I have to work with - 30 year old footage - and I can't do anything about the acquisition footage.

This video was likely shot with single chip, pre 1990 camcorder with cheap glass, using available florescent lighting with mixed tubes, auto everything, and at night with no hope of any light bleeding in from outside. I'd imagine the camera must have boosted the gain - a lot, just to get a murky image using the image sensor of that era, right? It probably would have looked a whole lot better using a professional three-chipper, good glass and lots of light.

I'll run some tests using Huffyuv 4.2.2 to see if it looks better than 4.2.0 or 4.1.1. and perhaps try another more modern Svideo to USB transfer device.

Thanks!
Steve

[WarbirdVideos]

@Hushpower: Good one! Yep - good ole Neat Video - I've had it for years. Throw in some Happy Otter for deinterlace, 60p conversion, upscale, and Y/C delay correction, Boris BCC Magic Sharp, some gamma correction, and you start to get something worthwhile! But that of course, comes after the best possible capture. Some day, with AI tools, these old videos will look like they were shot in 8k - or better!

[Hushpower]

The 710-USB is pretty close to as good as you're going to get, USB digitiser-wise. LS sold me one. I no longer use it because it was too temperamental with Win 10, but if you have it working, then stick with it.

Quote:

I'll run some tests using Huffyuv 4.2.2 to see if it looks better than 4.2.0 or 4.1.1.

The real test would be "how does it look after you have put it into it's final delivery format" eg H264/MP4.

[traal]

Post a short lossless clip going straight from the VHS VCR to the 710 with no postprocessing and we'll take a look.

But if you want the best possible quality, then you need a VCR with built-in TBC and also an external TBC from LordSmurf, with double shielded S-Video cables throughout.

[WarbirdVideos]

Will a TBC help the quality here? The video was shot in 1994 under florescent lighting, probably with boosted gain. I have no idea what camera was used, but color is usually dull in low light.

I used an S-video connection from a JVC HR S4600u, through a Panasonic DMR-HS2, into a Pinnacle 710 and Virtualdub. I have 65 VHS & Hi8 tapes to transfer, plus other miscellaneous Hi8 tapes shot outdoors with proper light.

Should I use a TBC during capture before post processing? If so, where can I get a TBC, and how much? Or, go with post processing using in Vegas 20, Neat video, and Happy Otter Scripting for video for deinterlacing and up-scaling to 60p? I want the best quality, so will a $2,000 TBC be best to start with?
Thanks,
Steve

[mbassiouny]

I see no one got back to your question, but if you still need an answer:

>Will a TBC help the quality here?
The quality of what?

There is signal quality and visual quality.
The improvement you gain with Frame TBC (AKA 2000$ TBC) is not (usually) visible to the naked eye, it does not improve visual quality (which is what you are complaining about).
Some of the cases where Frame TBC gives a noticeable difference is: Jitter correction, or some tapes may give you black and white for a few seconds when the signal is bad, but with a TBC the colors are restored, and some other signal correction that may or MAY NOT result in a visible improvement.

What improves visual quality is line TBC with 3D NR, This a VCR feature (usually) and can be found in many High end jvc decks, your 4600U is not one of them.
Some Line TBC-ish correction can also be achieved with passthrough devices such as the Panasonic DMR ES10.


As for DMR-HS2, I don't know that device and I cannot comment on its similarities to ES10.

> Hi8 tapes to transfer

Most Hi8 tapes I have encountered have been a PAIN! they usually require a Frame TBC, the pinnacle and hauppauge USB live 2 will give some random colored lines that barely resemble the faces of the humans in the videos.(at least this was my experience), in many cases, you can get something watchable with a Canopus ADVC, better than nothing at all with a pinnace, but with a Frame TBC and lossless capture card the quality is better.

[timtape]

Quote:

Originally Posted by WarbirdVideos

...Attached are still photos of the Pinnacle 710 and the DVCAM capture through VituralDub...

The two stills you posted contain virtually the same picture information as each other. Color balance and gain are virtually identical. The images only present differently due to some difference in brightness and contrast which can easily be adjusted in post, revealing two almost identical images, both to the eye and on the video scopes.