Proper way to capture composite output?
 

Hello all,
I have a single betamax tape I need to transfer. The player (Sony SL-HF300) only has composite out and is connected to my AVT-8710 (Green). I’m able to display the signal from the TBC S-video output. Or should I be capturing from composite? Capture device is an ATI600 USB 2.0. Any advice is greatly appreciated.

I'd capture from the S-Video output of the AVT-8710. The purpose of the frame TBC is to store a few frames digitally on a rolling basis. This acts as a frame buffer so that the frames can be released at a digitally precise rate and converted back to analog on the way out.

I maintain that the best output would be if you had access to the digital frames within the TBC and skip the analog conversion on the way out which then has to be re-digitized again later. This is possible for TBCs that have SDI or HDMI output that is still 480i. None of the "recommended TBCs" have digital outputs though.

For analog video, generally, the more wires that analog data travels over, the more separation and reduced interference between brightness/color/colors and more accurate reproduction of the starting image when the display device takes those signals to reconstruct the original image.

Hierarchy of analog signal quality usually is:
Worst = Composite (1 wire)
Medium = S-Video (2 wires - color and brightness separated)
Best = Component (3 wires - Brightness and difference between brightness and Blue and difference between brightness and Red), analog RGB/VGA are also in this category


This forum tends not to like component capture for some reason - probably because the recommended TBCs don't have it as an output and the recommended capture cards don't have it as an input. I've seen quite good results with an FA-310 which does have component out and you can feed it either composite or S-Video in. There are relatively few formats that are stored as component video, professional Betacam is the only one I can really think of.

Use crossover. Composite in, s-video out. Some TBCs do not allow this, but the AVT-8710 does.

This is videophile-level anal retentiveness. This loss is vastly overblown, essentially imperceptible to even graphs and meters (ie, beyond what the human eye can perceive).

And when you force digital out, you run into limitations on your workflow, which is why this wasn't done to begin with. You also overly rely on a single device ("one ring to rule them all!"), which never works out well. Manufacturers always screw up something.

This is a technobabbly statement. Wires don't cause bright/color shifts.

Correct.

No, wrong. The video source is Y/C, or Y+C, where luma is separate from both chroma. When you try to sub-separate both chroma, you have to process it. You don't want to do that in the analog hardware ingest workflow.

As per above, not the reason. :no2:

Conversations like this are screwy. You worry about a 0.0x% loss from TBC output, but then are willing to incur a big % loss from overprocessing the chroma? That makes no sense.

And thus why those rackmount TBCs were made for non-consumer (non-VHS) type sources. :wink2:

I think component being better for VHS is a pretty common misconception. I had seen this https://youtu.be/zB42-A5VOsQ?si=xhQa5NutAfl7hBuV where he is trying to argue that point. Then it goes into blackmagic and decode.

Yeah, that video came up on my feed some days ago. It's just 40 minutes of a guy rambling, mostly about white papers, Blackmagic, component. etc.

Certain orgs -- and their white papers / docs -- should be completely ignored. For example, AMIA docs were clearly written with a bias to pro media, and yet others constantly try to apply it to consumer media. The main author is/was an Ampex engineer, worked with Beta, etc. So nothing to do with VHS, Hi8, etc. In fact, in AMIA's various docs over the years, they crap all over S-VHS and Hi8 (not even mentioning lower quality VHS and Video8). I've had to rescue more than one org/library that misunderstood docs, and followed the bad advice (as it applied to VHS). They were NOT pleased with the quality they were getting, due to random reading, and trying to random apply it to their situation. Worse yet, the AMIA past docs have fearmongering that make it seem as if TBCs are bad.

Anyway, yep, component is misunderstood by video novices.

On video game consoles, I think S-Video color is slightly muted compared to RGB (i.e. RGsB/RGBS, not to be confused with component video's YPrPb), but only because the 3 native RGB signals (not including syncs) are being shoehorned into 2 wires (not including ground). VHS is already basically 2 signals (Y/C) and so the extra color information (gamut?) is already lost, and decoding/re-encoding to component video won't bring it back again.

DVD players use component video output for progressive scan, but that doesn't apply to VHS or Betamax.

Interesting points.

I don't think I said anything about bright/color shifts over wires. The more "separated" a signal is, then the less (or absence of) crosstalk you'll get between the components going over each wire. That is the only reason that S-Video is superior to composite - no crosstalk/overlap of the chroma and luma signals that can result in dot crawl or other displayed artifacts where certain patterns of luma data are incorrectly interpreted as chroma data and vice versa. This also results in a sharper overall image in most cases.

I think what is being missed here is that once you've run the analog signal through a TBC, the output from the TBC is based on the digital buffer frames created by the TBC, which are stored on a rolling basis as digital RGB data. For the most accurate TBC output, you want your output to be as close as possible to the digital RGB buffer frames.

Component/RGB is absolutely better than YC for representing any digital image or digital buffer frame. Better yet is just saving the digital source buffer frames as they are, which is what you get with SDI.

I'm also not sure what is meant by "chroma over processing" with component. It takes more "processing" to go from a TBC's RGB digital buffer to YC (data gets blended together) than it does to go from digital RGB to analog Component.

To clarify, I'm not saying component or SDI is better to capture VHS in general - I'm saying it's the superior option to capture from a frame TBC if it has those as output options specifically. S-Video would still ideally feed the TBC. If you aren't using a frame TBC, I'm not suggesting converting to component or SDI first and then capturing it. That could still be a viable option, it's just hard to tell if frames are being dropped or duplicated within the conversion device.

Now, whether the average human can tell the difference between YC/component/SDI captured off of a frame TBC for vintage video sources in most cases is a completely different question. ....However, if you had access to all 4 simultaneous outputs from a TBC (SDI, Component, S-Video (YC), Composite), why not capture off of the one that is going to most accurately represent the digital buffer frames that the output is directly based on?

TBCs do not perform a color conversion. If a TBC converted to RGB it would defeat the purpose of a proc amp because a proc amp has to go after a TBC to prevent baking in timing errors and it needs to come before the RGB conversion to prevent clipping.

VHS is stored in YPbPr. Component is YCbCr. When you use component you are converting to YCbCr and you’re doing the conversion in the analog domain.

That is incorrect, Consumer analog tapes are Y-C, Some technical papers call it composite format because the chroma is encoded into the luma at the RF level, YPbPr is the analog video component, YCbCr is the digital video component.

Warning: This thread is getting far removed from the OP question. :depressed:

However...

Well, no, an important distinction needs to be made here. It's not really "the wire", but rather the signal output. Wires are dumb, and carry whatever you give it. Length, shielding, impedance, determines continuity and integrity, but that doesn't alter colors. Mere wire don't know luma from chroma, nor care, nor need to care. I don't think there's actually a way to transmit RGB out, and have a Y/C (YUV) display understand it. So the RGB is rough converted to YUV at the console output. Nothing to do with wires, or even s-video itself. It's all about the console processing.

That same concept is why composite mostly sucks. It's not actually "composite" (composited Y+C) doing it, but rather lousy processing.

Game consoles have to be signal tapped before that processing happens, in order to extract the raw(er) sharp(er) original digital quality.

This is not true. When you bifurcate merged chroma, the processing to achieve this is not transparent. See also, the difference between 2D and 3D sampling.

What you say is only true if the origin signal is separated already, a true Y+U+V, and almost always a digital signal (not analog). Ignoring displays/TVs/HDTVs, there really is not a true component analog source, aside from Betacam (and even that implementation had compromises).

No, not correct, it's never RGB data.

Component isn't necessarily RGB (and in fact almost never is), especially not anything plugged into your HDTV (YUV).

This statement has compounded factual errors, and thus comes across as technobabble.

Well, good, because it's not.

But it's not something that exists in a practical reality. (Yes, you have SDI devices that "all in one" the process, but when has any AIO solution not made compromises? It's essentially the "one size fits all" of video ingest. It can fit many, or it can fit none.)

And thus one of the concerns of closed-loop workflows. The magic box does it all for you, you know nothing (and don't need to know, according to the device manufacturer). "Just trust us!" (If you're a fan of the Slayer anime show, "it's a secret!".)

I think you're making an assumption here, and I would wager it's precisely because of not having used any of these methods for a sustained period.

The enclosed/AIO SDI workflows are not better. Those are merely different to achieve (hopefully) comparable results. The problem comes when tapes balk at it, and you require an entirely different setup. With a more modular traditional setup, that never happens. Tapes can, and do, balk at everything: VCRs, TBCs, capture cards, and optional devices like detailers or proc amps. When the items are separate, you can subtractive troubleshoot. When you have a single box, problem solved, "it doesn't work, use something else".

I have SDI boxes. I don't often use them.
I have component, HDMI, etc gear. I rarely to never use any of it. It was acquired for R&D. I never shared, I never had time to. But moving forward, I want to, but it will require that I tagteam work with other members here, and I've slowly been reaching out to some of them.

I think this is a case of alphabet soup semantics. Y+PbPr is probably more accurate.

Watch out using the word "component" when a non-jargon synonym(ish) will do. :laugh:
Better = "YPbPr is the analog video reference, YCbCr is the digital video reference".