Maximum/Highest Quality Signal Transmittable Over S-Video?

[mrmuy97]

Have to start this first post with the obligatory massive thanks to everyone contributing to these forums. Special thanks of course to Lord Smurf and several other notable members whose excellent posts I continually come across.

I'm trying to learn as much as possible about all this, but am having a hard time finding info on what the highest-quality signal that can be carried over S-Video is, if it's even possible to measure in analog terms, or digital equivalent, or anything.

I'm interested in that because one setup I'm thinking of would capture the S-Video output at up to 10-bit 4:2:2 uncompressed. It seems like that should yield a visual improvement over 8-bit 4:1:1 (or 4:2:2 if it's an option) uncompressed using the same capture device. Specifically I read that "10-bit represents a significantly higher quality," and "because 10-bit video has four times the numerical precision when compared to 8-bit, it has a signal-to-noise ratio 12 db higher than 8-bit video." But I'm uncertain as to whether there's enough info delivered over S-Video in the first place to make any difference between 8- or 10-bit capture, and perhaps it would need to be component or better to matter. Or if maybe the signal over S-Video is irrelevant and the 10-bit capture & digital conversion will always be superior to the 8-bit, no matter what. Real-world experience to back any of the theoretical stuff up would be great. Hopefully I'm making some sense here haha.

I found one reference in "Digital Signal Processing 101" by Michael Parker which stated that S-Video "carries a 4:2:2 YCrCb signal in single form over a multipin connector." However, my understanding is that Y/C being analog means it can't be 4:2:2 at the same time since that's a digital measurement, which leaves me questioning whether that's an accurate statement.

I have to think that the 10-bit 4:2:2 uncompressed capture/conversion is approaching (if not already past) a point of extremely diminishing returns in this case. My source is a collection of home movies from childhood, sadly all shot on Memorex, Sony, RCA, Maxell, and other assorted tapes which appear to be of known poor quality, but at least all recorded on the same Panasonic PV-420D, always T-120s in SP mode, and mostly only single-use and not over-watched afterward, stored well, etc. The associated size, bandwidth, storage, etc, are acceptable to me for this project if 10-bit results in a reasonable improvement versus 8-bit. This week I acquired a good S9500U, and I may be working out a deal for a 1980 with a dead front LCD display and a functional HD2000U. Even if the tapes turn out to be particularly low quality, I want to get the best capture as is reasonably possible.

Now I've managed to make this much longer and drawn-out, and I may or may not have even properly asked what I'm trying to figure out, but in any case I appreciate any and all help/info/etc. Hopefully at some point I'll have enough knowledge and experience to help others as well.

[lordsmurf]

Historical reference may matter here.

"s-video" (separated video) and S-VHS (Super VHS) came out at the same time, both released in 1987 by JVC. I don't know if JVC invented the technology, or simply bought/licensed it from another company. Many falsely believe it to be "super video", and that it was made for S-VHS, but that's not accurate. They merely needed a connection that would show the full resolution of the format, and both RF/coax and composite were not it.

It should be mentioned that SCART had already existed since the 1970s. And its RGB signal (aka interlaced component) was better than s-video. But sadly it was not used.

VHS formats (and thus S-VHS) used "4:2:2" color. Yeah, that's not entirely accurate, since I'm foisting digital terms onto analog. The chroma was half the luma value. VHS luma was a miserable "240 lines", and by the time you convert it to digital numbers, factoring in the 50% color, it's about 240x480.

S-VHS came along and almost doubled the luma. In false digital terms, that's like saying "7:2:2". Great resolution, but crappy color. It was about "400 lines" analog, and about 400x480 digital.

VHS color wasn't even 8-bit. Capturing or processing it 10-bit is overkill.

Note: The numbers you read on Wikipedia (and other places) are all BS, only consider the luma, and use the disputed Kell factor in the equivalency. In other words, theory (on paper), not in practice (reality). Bad theory, at that! I don't care about theory anymore than I care about unicorns and UFOs.

The tape quality and recording mode really affected quality as well.

Anyway, that's why it was made -- to get (at minimum), in digital terms, a 400x480 resolution.

At best, s-video cables will transmit about 550x480 resolution. Some claim 640, but I disagree. Some even claim 720x480, but they confuse that with the upper limit of SD (standard definition) video. Even "SD" TVs never really got above 550x480, and that was only on high-quality SD CRT sets made during the HD days. I have one from JVC.

We often transfer DV (720x480) via s-video. Oh no, you say! We lose the resolution! Not really. The cheap glass on the camcorder already did that. And that's not even considering the overall resolution lost from NTSC 4:1:1 or PAL 4:2:0, which raped the chroma compared to a 4:2:2 analog tape. Only when a professional DV SLR camera was used do we bother with the actual Firewire transfer.

As far as 8-bit vs 10-bit s-video ... ehh. I think that a full 4:4:4 type resolution is possible over s-video (not mere 4:2:2), meaning that 10-bit may make a small difference -- but only if the source is 4:4:4 (or at least something high-end like Betacam). It will make zero difference on VHS or Video8 formats.

Your biggest issue is not the resolution -- it's the chroma:luma overlay on the tapes. Avisynth is usually needed.

Does this all make sense?

Fun conversation.

[themaster1]

If you want to enhance an analog signal you need the right tools (and they're expensive) example [/url=http://www.cypconverters.com.au/digital-mixers-base-correctors/ctb-100.html]here[/url]
Then the quality of vcr, quality of cable, radio/emi interferences nearby, the capture card will come into play.
Most "cheap" capture cards capture ycrcb 4.2.2 and that's fine for vhs/home movies

[msgohan]

I've disputed some of the numbers listed above by lordsmurf in the past, with visual evidence, but it's fallen on deaf ears. I should probably leave it at that. I agree with his conclusion that you shouldn't bother trying to capture 10-bit.

Your thread topic question is irrelevant since you're wanting to capture VHS. The quality of the interface far surpasses the quality of the medium.

Have you worked with 10-bit video before? The Avisynth workflows aren't for the faint of heart.

[lordsmurf]

Quote:

Originally Posted by msgohan

I've disputed some of the numbers listed above by lordsmurf in the past, with visual evidence, but it's fallen on deaf ears.

No, just bad memory. I'd very much like to re-look at your findings.

[mrmuy97]

@lordsmurf:

Ha, good old RF/coax... I remember being ~5 years old and popping Zelda into my new NES and even then noticing the poor picture quality showing up on that wonderful 13" Zenith. The difference between composite and S-video wasn't apparent to me as quickly as when switching from coax to composite (probably because the average TV display was still awful anyway), and I don't think I really paid attention to S-video until it became an output option on graphics cards. Now I can hook up the Genesis with composite or S-video and see a night-and-day difference, though now it's on an LCD display bearing no resemblance to the TVs of that time. I remember reading about SCART and wondering why it was decided that VGA/RGB/D-sub would be great for computers but not get similarly used here for other A/V equipment. Now there's people like the guy who made little boards for the N64 to bypass the original video output circuitry by soldering to the RCP and using RGB out for the first revision, then DVI, and now Mini-HDMI with improvements each time. Granted that is working with a digital source and just figuring out how to output it better, but it makes me wonder if any such improvements could be made for a widely-used deck like the 1980 to grab the signal before getting merged and output via the Y/C interface to instead use component output like the W5U, or HDMI like the N64 mod.

Thanks for the practical digital and analog info. The 4:2:2 analog approximation is one of the things I was particularly curious about, as well as if the 256 vs 1024 difference with 8- vs 10-bit would matter at all considering, like you said, VHS' already "miserable" color. I'll never forget watching just about any "home movie" on VHS and realizing how terrible the picture was, mostly in terms of color. After reading through a small portion of the knowledge here I finally understand why, as well as why I never noticed such glaring issues when watching commercially-made tapes like Star Wars. Those things probably looked like Blu-rays to me compared to the garbage created by consumer equipment.

I've read reports of using the DVX-1000A with the "MiniDV TBC/ADC combo" method to do the A/D conversion at 12-bits and 4:2:2 into DVCPRO50, but that's an expensive piece at ~$300-$400 minimum used, and 12-bit definitely seemed excessive, plus still getting compressed to DV, even though it remains 4:2:2 rather than 4:1:1 as with the usual 8-bit ADC MiniDV cams. It's good to know that the 12- and 10-bit options will not yield any discernible difference compared to 8-bit. There's no way I'd be dishing out hundreds for something like a DVX-1000A to use for conversion anyway, unless there was a substantial improvement to be had.

I've never used Avisynth but it is indeed on my list of things I'll most likely need to become familiar with.

Regarding 4:2:2 vs 4:1:1, I found this guy's work interesting. [http://www.nattress.com/Chroma_Investigation/chromasampling.htm] He was able to get his 4:1:1 video much closer to 4:2:2 than I thought would be possible. But I still like the idea of having uncompressed 4:2:2 masters to archive and probably x264 versions to keep readily available for viewing, sharing, etc. I know I will also have to try the MiniDV method myself so I can see how it compares to the uncompressed 4:2:2 capture from the deck's S-video. Also glad to hear that you have faith in S-video's data capacity.

@themaster1:

I don't think any external TBC hardware is an option for my budget if it's all hundreds or thousands of dollars as is often brought up. The best that I can get with the deck's internal TBC and editing software will have to suffice. Most, if not all, of these tapes are first-generation and have not been played excessively, so I'm hoping to not have a terrible time with them. The quality of the entire signal chain from the physical tape to the digital conversion is indeed my highest priority.


Now I can't stop thinking about a modded 1980 with the chip(s) that handle all the DNR, TBC, ADC, etc in a MiniDV cam added inside the deck along with an on/off switch just like the deck's internal features, and that MiniDV's ADC output tapped and sent via component or HDMI. Mmmmmm. But that seems too good to be possible, and surely the MiniDV's ADC converts straight to the DV format anyway, compression and all. Oh well, wishful thinking.

[mrmuy97]

@msgohan:

I hope to see you and lordsmurf revisit your findings and see what conclusions you both draw. At least for now you are both in agreement as far as using 10-bit for VHS. I have not worked with 10-bit, which led me to ask about S-video's capability. Perhaps a more accurate phrasing would have been to ask whether the VHS source is capable of being read and sent in high enough quality via S-video that an uncompressed 10- vs 8-bit capture/conversion would be beneficial.

An additional question that may be getting rolled into this unintentionally would be whether the same initial signal is improved when sent from the deck over component as YCbCr rather than via S-video as Y/C, considering the source's limited quality. I'd love to know if anyone with a W5U or other deck with factory component output has compared that against the S-video output. Is there anything left to gain if VHS used approximately 4:2:2 and S-video is already transmitting the same?

[lordsmurf]

I still have a 13" CRT from the 90s, and used it in my kitchen before my move. I often referred to it as my "video filter" for poor quality formats like VCD or unrestorable VHS. It was so small, and so undetailed, that it made viewing such content a not-miserable experience. Funny but true!

I sometimes think Europe was treated like a step-child when it came to video, in the analog days. Part of it was their own doing, having so many competing formats for relatively small countries. In-fighting on the pre-EU days is to blame. Thankfully that has all changed now in the digital and HD eras.

DV is really just ancient now, a product of the late 90s / Pentium III era. It was a created primarily to save space, and quality was secondary. There's no reason to use it now. I now have thumb drives that are bigger than every computer I owned in the 80s and 90s combined. I have about 50tb of external storage here -- and it all fits in a desk drawer.

Unfortunately, that guy's site has images that are too compressed (JPEG) and too small (must be downres'd) to be overly useful. You'll just have to take his word for things.

You need a TBC. Get a DataVideo for about $150-225 off eBay.

I think the main difference between msgohan and myself is he's using luma only to measure resolution. I've always factored in the resolution loss for a "bigger picture" equivalency. The color-under of VHS makes it less than that of 4:2:0 DVD, even though it's still a "4:2:2" when cramming VHS into digital world terms. Color affects overall resolution, which is why Kell is ridiculous, and why I don't just use B&W charts. But, yes, I want to revisit his research. I may redo our resolution guides to include both aspects.

You can't output VHS as component anyway, not capture VHS as RGB SCART. s-video is the only option.

Some members here have a JVC W5U. I suggest opening a new thread so it's seen.

[msgohan]

This is the post to which I never received a reply. We can discuss resolution further in that thread if you'd like.

@mrmuy97, there are some VCRs with HDMI outputs if you are really interested, but most of them are just VHS/DVD recorder combo units. Exceptions: the HM-DH5U and HM-DT100U, JVC's only D-VHS models with HDMI; they should output a similar picture to their S-VHS units that are on the buying guide. Recent eBay prices range from $250-$450 according to a completed listings search.

But if you play one tape in multiple VCRs, the quality variation you see from each model's processing will be much larger than the difference between using various outputs on a given model. That's even true about composite vs S-Video in most cases.

Additionally, HDMI output isn't all it's cracked up to be when the original signal is analog. VCRs don't offer Proc Amp controls, so you're limited to working with whatever dynamic range and colors the analog-to-digital converter inside gives you. If there is clipping in any channel, there is nothing you can do to recover the signal.

[NJRoadfan]

S-Video or Y/C separated video existed well before SVHS decks arrived, the Commodore 64 computer outputted it and Commodore even offered a monitor with the input via dual RCA jacks. The resolution increase is noticeable if you have a source and monitor that can take advantage of it. SVHS utilized 7Mhz of luminance resolution while ED Beta put out a staggering 9Mhz. Both formats retained the sub 1Mhz chroma bandwidth however.

You won't ever see that level of detail on broadcast NTSC signals though since they were limited to about 5Mhz luma and 1.5Mhz chroma (total NTSC-M channel bandwidth is 6Mhz).

[mrmuy97]

Turns out the simple answers to some of my questions were already posted in the "Guide to Understanding Video Sources, Part 2 – Capturing Videotapes" [http://www.digitalfaq.com/guides/vid...nd-sources.htm]. Under "Analaog Source Resolutions" there's a table citing "Analog Measurement" and "Digital Equivalent" as well as other relevant info from LordSmurf (also posted by him in this thread) which should be of interest to anyone who might come across this thread searching for similar info. I should have searched a bit harder before posting and clarified the question(s).

Quote:

Originally Posted by msgohan

But if you play one tape in multiple VCRs, the quality variation you see from each model's processing will be much larger than the difference between using various outputs on a given model. That's even true about composite vs S-Video in most cases.

Thanks, that makes sense, and having multiple decks to test each tape in is some of the most common advice I've seen, clearly for good reason.

Sounds like 8-bit 4:2:2 over s-video should be the first thing to try when I get to capturing. Until then I must continue reading.

Quote:

Originally Posted by lordsmurf

VHS formats (and thus S-VHS) used "4:2:2" color. Yeah, that's not entirely accurate, since I'm foisting digital terms onto analog. The chroma was half the luma value. VHS luma was a miserable "240 lines", and by the time you convert it to digital numbers, factoring in the 50% color, it's about 240x480.

VHS color wasn't even 8-bit. Capturing or processing it 10-bit is overkill.

At best, s-video cables will transmit about 550x480 [VHS?] resolution. Some claim 640, but I disagree. Some even claim 720x480, but they confuse that with the upper limit of SD (standard definition) video. Even "SD" TVs never really got above 550x480, and that was only on high-quality SD CRT sets made during the HD days. I have one from JVC.

As far as 8-bit vs 10-bit s-video ... ehh. I think that a full 4:4:4 type resolution is possible over s-video (not mere 4:2:2), meaning that 10-bit may make a small difference -- but only if the source is 4:4:4 (or at least something high-end like Betacam). It will make zero difference on VHS or Video8 formats.

Your biggest issue is not the resolution -- it's the chroma:luma overlay on the tapes. Avisynth is usually needed.

Quote:

Originally Posted by msgohan

I agree with his conclusion that you shouldn't bother trying to capture 10-bit.

Your thread topic question is irrelevant since you're wanting to capture VHS. The quality of the interface far surpasses the quality of the medium.

[mrmuy97]

7 years later...

As of yesterday I finally had the bare minimum equipment needed (deck and capture card), so I took my first test capture last night. HS-HD2000U s-video output >> Extron s-video to BNC Y/C split cable >> Kona LHe breakout cable BNC Y/C inputs >> capturing 8-bit 4:2:2 uncompressed in Premiere Pro CS6. I think CS6 is the last version that supported my LHe. About 1GB per minute and I'm pleasantly surprised by how it came out. I tested on a THX Return of the Jedi VHS that seems to be in good condition. x265 compressed it to about 1.4% of the original uncompressed size with almost no quality change. Excited to keep going and start trying different settings and eventually be ready to do the home videos properly.

This setup is temporary as I just couldn't wait to get my first capture by any means necessary. I know none of this is the recommended stuff but it's just what I already have set up and am already familiar with using. After I get a TBC I'll be ready to go for real and I'll have to move on to learning about the capturing software.

[latreche34]

You waited 7 years for the Aja Kona? Wow, Anyway I'm glad you settled on a good workflow.

[mrmuy97]

Quote:

Originally Posted by latreche34

You waited 7 years for the Aja Kona?

Ha, of course the long wait was for the right VCR. Wanted one with minimal use and from the original owner in original box, all accessories, mint condition, etc. Took a long time. I had a JVC S9500U which worked briefly and then started eating tapes as soon as it began to play them. Normally I fix/maintain my own equipment but I decided in that case to go with a completely different unit. S9500U does make a nice "retro" decoration in the entertainment area now though That was $250 in 2015 so it's not too terrible a sunk cost at the end of the day.

Although I don't recall there being many LHe cards on eBay back then when I was thinking of trying one out, I was able to purchase one from the original owner when he upgraded his capture equipment to the LHe Plus. Received it in its anti-static bag with breakout cable in its original box for $40 back in 2015. So I've been sitting on that this whole time.

I see now the LHe is going for like $30 and there are a ton of them available. Good price in my opinion, surely it's at least better than most of the common Chinese crap out there now. Pair with an Extron s-video Y/C split cable and you're in business.

Now I'm working on a TBC, but I've only come across 1 reliable source for those (LS), short of these very rare eBay unicorn units. Getting the best that the budget allows.

[latreche34]

There is a member here who used the 12G version of the card and he demonstarated it with pretty lousy VCR's and tapes and produced pretty stable images, Do more experimenting with yours before you go out and dump more money, You never know what's under the hood of those high quality cards.

[mrmuy97]

Quote:

Originally Posted by latreche34

There is a member here who used the 12G version of the card and he demonstarated it with pretty lousy VCR's and tapes and produced pretty stable images, Do more experimenting with yours before you go out and dump more money, You never know what's under the hood of those high quality cards.

Any chance you have that thread saved or can remember what it is? Just looked up that card, pricey haha.

[mrmuy97]

I was just reading on YUV info and came across another agreement on sticking to 8-bit maximum for VHS:

8- and 10-bit sampling.
In principle, 10-bit encoding is superior in clarity to 8-bit, due to the reduction in tonal contouring and other artifacts. Some specialists argue, however, that there is no benefit for certain classes of material. One university expert wrote, "We digitize Betacam SX tape to 8-bit UYVY but Digibeta to 10-bit V210 because these selections align with the nature of the data that is actually sent out over SDI from these tapes. . . . SDI is 10-bit data, but when I piped the SDI video data from an SX tape to a binary display I could see the 9th and 10th bits were always zero. Thus by taking only the first 8 bits I could get all meaningful data . . . . I have about 3,000 SX hours to preserve and choosing 8-bit instead of 10 saves me about 90 TB of storage"

https://www.loc.gov/preservation/dig...dd000365.shtml

Quote:

Originally Posted by lordsmurf

VHS color wasn't even 8-bit. Capturing or processing it 10-bit is overkill.

As far as 8-bit vs 10-bit s-video ... ehh. I think that a full 4:4:4 type resolution is possible over s-video (not mere 4:2:2), meaning that 10-bit may make a small difference -- but only if the source is 4:4:4 (or at least something high-end like Betacam). It will make zero difference on VHS or Video8 formats.

Your biggest issue is not the resolution -- it's the chroma:luma overlay on the tapes. Avisynth is usually needed.

Quote:

Originally Posted by msgohan

I agree with his conclusion that you shouldn't bother trying to capture 10-bit.

Quote:

Originally Posted by mrmuy97

Specifically I read that "10-bit represents a significantly higher quality," and "because 10-bit video has four times the numerical precision when compared to 8-bit, it has a signal-to-noise ratio 12 db higher than 8-bit video." But I'm uncertain as to whether there's enough info delivered over S-Video in the first place to make any difference between 8- or 10-bit capture, and perhaps it would need to be component or better to matter. Or if maybe the signal over S-Video is irrelevant and the 10-bit capture & digital conversion will always be superior to the 8-bit, no matter what. Real-world experience to back any of the theoretical stuff up would be great.

I have to think that the 10-bit 4:2:2 uncompressed capture/conversion is approaching (if not already past) a point of extremely diminishing returns in this case.

[latreche34]

Most script software don't work in 10bit anyway.

[traal]

10 bits of dynamic range (luma) would be nice so I wouldn't have to worry so much about setting the brightness and contrast settings correctly in VirtualDub.