Newbie setup for VHS capture

[zibon]

Hello, I am new to OBS and the forum. I am attempting to capture some VHS and have questions about the settings. I am using a 1080p upscaler into a USB3 capture device. I have a 2020 M1 Mac Mini.

I have successfully recorded audio and video. The video is fairly blocky. I've been referencing many various online suggestions for settings and they're a bit all over the place. Video is set to 1920x1080 (base & output) and 59.94 fps. Also in the capture card properties deinterlacing is set to Yadif 2x. I've attached the output settings.

The goal is to have the VHS captures available as files. I am doing this for someone else, so I selected mp4 for compatibility. Any suggestions with other settings (bitrate, encoder, etc) to get the best looking output? If file sizes need to be large for the best capture, I have experience editing and compressing. Thank you!

 

[AaronD]

I am using a 1080p upscaler into a USB3 capture device.

It's not a cheap USB capture, is it? Those are deceptively-marketed garbage.

Perhaps the most blatant deception is a cheap USB 2 chip behind a genuine USB 3 connector (the additional pins for USB 3 are indeed present, but not connected), and advertised as USB 3 with all of those benefits, when that is absolutely false.

That then, leads directly to a practical problem of picture quality. You can't cram HD video through USB 2 (which it actually is, despite all the hype about USB 3), and so it has to compress it, hard, in the card itself before it even gets to the computer at all. And it uses a dirt cheap method to do it too: MJPEG, which is just a JPG still image of each frame, with no knowledge of the other frames so as to take advantage of their similarity. So the quality is even worse than that bitrate might be otherwise.

And there's the usual haphazard design and/or manufacturing at such a low price point for something that usually costs several times that, which leads to random things not working or only partially working...and it's too cheap to be worth returning.

---

If you must capture true HD video on USB, stick with the name brands that actually care about customer loyalty, and thus have some accountability to those customers. And expect to pay around $80 to $120 for a single input.

But, since your original source is not HD, you don't need that. Nor do you need the upscaler. Get a good Composite and/or S-video to USB capture - those CAN work on USB 2, but still use the same caution about manufacturer accountability - and set OBS to use a suitable canvas and output size. Probably 640x480 at 59.94fps, and use the Yadif 2x deinterlacer.

(S-video is slightly better if your VCR supports it. Technically the same specs as Composite, but it separates the brightness, color, and sync signals so there's less interference/crosstalk between them.)

---

Also use MKV for recording, even if your finished file is going to be MP4. Then have OBS remux it afterwards. No loss of quality that way - it just shovels the data across exactly as-is, into a different container - but it makes the recording more resilient to crashes and other sudden stops. If you record directly to MP4 and something happens, you lose the entire file. If you record to MKV, you still have everything up to the crash point.

 

[zibon]

It's not a cheap USB capture, is it? Those are deceptively-marketed garbage.

Perhaps the most blatant deception is a cheap USB 2 chip behind a genuine USB 3 connector (the additional pins for USB 3 are indeed present, but not connected), and advertised as USB 3 with all of those benefits, when that is absolutely false.

That then, leads directly to a practical problem of picture quality. You can't cram HD video through USB 2 (which it actually is, despite all the hype about USB 3), and so it has to compress it, hard, in the card itself before it even gets to the computer at all. And it uses a dirt cheap method to do it too: MJPEG, which is just a JPG still image of each frame, with no knowledge of the other frames so as to take advantage of their similarity. So the quality is even worse than that bitrate might be otherwise.

And there's the usual haphazard design and/or manufacturing at such a low price point for something that usually costs several times that, which leads to random things not working or only partially working...and it's too cheap to be worth returning.

---

If you must capture true HD video on USB, stick with the name brands that actually care about customer loyalty, and thus have some accountability to those customers. And expect to pay around $80 to $120 for a single input.

But, since your original source is not HD, you don't need that. Nor do you need the upscaler. Get a good Composite and/or S-video to USB capture - those CAN work on USB 2, but still use the same caution about manufacturer accountability - and set OBS to use a suitable canvas and output size. Probably 640x480 at 59.94fps, and use the Yadif 2x deinterlacer.

(S-video is slightly better if your VCR supports it. Technically the same specs as Composite, but it separates the brightness, color, and sync signals so there's less interference/crosstalk between them.)

---

Also use MKV for recording, even if your finished file is going to be MP4. Then have OBS remux it afterwards. No loss of quality that way - it just shovels the data across exactly as-is, into a different container - but it makes the recording more resilient to crashes and other sudden stops. If you record directly to MP4 and something happens, you lose the entire file. If you record to MKV, you still have everything up to the crash point.

Thank you very much for the reply. The capture device is a Guermok. It’s HDMI and I already owned an upscaler, which is why I was going this route. And I also watched an in depth video of the process and the Guermok was one of the suggested capture devices (yes, could’ve been a sponsor, I don’t recall).

Here is the product site. I don’t know if you can tell if it is any good. I have gotten prompt and thorough replies from customer service I will add.

GUERMOK Video Capture Card, USB3.0 HDMI to USBC Capture Card, 4K Full 1080P60FPS Capture Devices for Gaming Live Streaming, Capture Video for PC Mac OS System OBS Zoom PS4/5 Switch Xbox1 (Gray)

【New Chip, Faster Speed】 This HDMI game capture card uses a newest chip technology, lower latency compared to the old model, faster transmission speed, better meet the needs of high frame rate video capture. USB3.0 high-speed transmission interface, support 4K 30fps input, perfectly adapted to...

www.guermok.com

My VCR does have S-video. If I end up getting another capture device, do you have a recommendation? And if the end result is that my friend will be probably casting these files to a tv, should I leave them at 640x480? I was thinking to avoid him having to make adjustments on his end and just have these files already upscaled. If it’s ok to upscale, should I do that post capture? Thanks again. I really appreciate it.

 

[AaronD]

Thank you very much for the reply. The capture device is a Guermok. It’s HDMI and I already owned an upscaler, which is why I was going this route. And I also watched an in depth video of the process and the Guermok was one of the suggested capture devices (yes, could’ve been a sponsor, I don’t recall).

Here is the product site. I don’t know if you can tell if it is any good. I have gotten prompt and thorough replies from customer service I will add.

GUERMOK Video Capture Card, USB3.0 HDMI to USBC Capture Card, 4K Full 1080P60FPS Capture Devices for Gaming Live Streaming, Capture Video for PC Mac OS System OBS Zoom PS4/5 Switch Xbox1 (Gray)

【New Chip, Faster Speed】 This HDMI game capture card uses a newest chip technology, lower latency compared to the old model, faster transmission speed, better meet the needs of high frame rate video capture. USB3.0 high-speed transmission interface, support 4K 30fps input, perfectly adapted to...

www.guermok.com

Not a brand I've heard of before, and the price says "cheap garbage". If they're selling them at a loss, then it might make sense...except for the loss on their end.

My VCR does have S-video. If I end up getting another capture device, do you have a recommendation?

Not really. I got mine from a friend at work who had moved to other things, and it "just happens" to work for me. Don't know what his process was to pick that one, or whether it's actually good or just works by accident.

But if you're buying one yourself, do remember what I said above:

• Stay away from the cheap ones.
• Research some brand names, by name, to see which ones really care and which are more like well-spoken scammers.

And if the end result is that my friend will be probably casting these files to a tv, should I leave them at 640x480? I was thinking to avoid him having to make adjustments on his end and just have these files already upscaled. If it’s ok to upscale, should I do that post capture?

The signal coming out of the VCR has about that many visible lines and no real horizontal resolution. To make the pixels square, you end up pretty close to 640x480, and that may in fact be what the capture device gives you anyway.

You're not adding anything by upscaling, though some upscalers look better than others. If it's displayed on a bigger screen, then there *will* be upscaling, *somewhere*. TV's are usually pretty good at it, but if you feed the TV from a computer, then it's actually the computer that upscales and not the TV...
The decision of whether to upscale yourself or not, depends on how much you trust your friend's rig to do it well, and on storage space. The original resolution will be a much smaller file for the same content.

And yes, I would record raw to a file, probably "Indistinguishable" or "Lossless", and then post-process that file if needed. It may or may not be needed.

Try a few, "Indistinguishable", on your friend's rig or equivalent, and see. If that works, check the actual file size. If that's okay, then do them all like that and call it done. If something's wrong, record "Lossless" instead, post-process as needed in a video editor (not OBS), export that, and when you really are *completely* done, you can delete the intermediate files if you want.

 

[zibon]

Not a brand I've heard of before, and the price says "cheap garbage". If they're selling them at a loss, then it might make sense...except for the loss on their end.


Not really. I got mine from a friend at work who had moved to other things, and it "just happens" to work for me. Don't know what his process was to pick that one, or whether it's actually good or just works by accident.

But if you're buying one yourself, do remember what I said above:

• Stay away from the cheap ones.
• Research some brand names, by name, to see which ones really care and which are more like well-spoken scammers.

The signal coming out of the VCR has about that many visible lines and no real horizontal resolution. To make the pixels square, you end up pretty close to 640x480, and that may in fact be what the capture device gives you anyway.

You're not adding anything by upscaling, though some upscalers look better than others. If it's displayed on a bigger screen, then there *will* be upscaling, *somewhere*. TV's are usually pretty good at it, but if you feed the TV from a computer, then it's actually the computer that upscales and not the TV...
The decision of whether to upscale yourself or not, depends on how much you trust your friend's rig to do it well, and on storage space. The original resolution will be a much smaller file for the same content.

And yes, I would record raw to a file, probably "Indistinguishable" or "Lossless", and then post-process that file if needed. It may or may not be needed.

Try a few, "Indistinguishable", on your friend's rig or equivalent, and see. If that works, check the actual file size. If that's okay, then do them all like that and call it done. If something's wrong, record "Lossless" instead, post-process as needed in a video editor (not OBS), export that, and when you really are *completely* done, you can delete the intermediate files if you want.

Thank you again for the information!

 

[zibon]

Not a brand I've heard of before, and the price says "cheap garbage". If they're selling them at a loss, then it might make sense...except for the loss on their end.


Not really. I got mine from a friend at work who had moved to other things, and it "just happens" to work for me. Don't know what his process was to pick that one, or whether it's actually good or just works by accident.

But if you're buying one yourself, do remember what I said above:

• Stay away from the cheap ones.
• Research some brand names, by name, to see which ones really care and which are more like well-spoken scammers.

The signal coming out of the VCR has about that many visible lines and no real horizontal resolution. To make the pixels square, you end up pretty close to 640x480, and that may in fact be what the capture device gives you anyway.

You're not adding anything by upscaling, though some upscalers look better than others. If it's displayed on a bigger screen, then there *will* be upscaling, *somewhere*. TV's are usually pretty good at it, but if you feed the TV from a computer, then it's actually the computer that upscales and not the TV...
The decision of whether to upscale yourself or not, depends on how much you trust your friend's rig to do it well, and on storage space. The original resolution will be a much smaller file for the same content.

And yes, I would record raw to a file, probably "Indistinguishable" or "Lossless", and then post-process that file if needed. It may or may not be needed.

Try a few, "Indistinguishable", on your friend's rig or equivalent, and see. If that works, check the actual file size. If that's okay, then do them all like that and call it done. If something's wrong, record "Lossless" instead, post-process as needed in a video editor (not OBS), export that, and when you really are *completely* done, you can delete the intermediate files if you want.

Hello, I just wanted to follow up on this and ask a few questions after having done a lot of tests.

I am still using the same setup (because its what I have). My friend that I'm doing this for has a lot of tapes. So I want to hopefully keep these to one step - capture and that's it with no post precessing.

I have been capturing at both 640x480 and 720x480 and they seem the same quality-wise (I am casting these from my phone to my TV to check). The upscaling looks pretty good on my TV. I of course will have my friend check some sample clips on his. I'm fine going with 720 to make sure I get the entire image. Do you think there is an issue with leaving the black on the sides? I've read some people capture at 720 and crop later to 640. Again, I'm trying to just to do one capture pass and be done.

I've included my current settings, but not much has changed. I changed the bit rate and a few of the options after gathering an average from articles I've read. Please feel free to comment on any of that if you would have suggestions.

Also, as a reminder of my setup, although I am capturing at 720, I am still running into the HDMI upscaler, then out HDMI to the capture card. So I am upscaling to 1080 and capturing 720x480, which I may change. I am considering getting a used BlackMagic Intensity Shuttle so that I can use the S-video and also because this generally seemed to be rated as a great USB capture card. All that said, I still think the samples have looked fine when cast to my TV.

Thanks again for the help.

 

[AaronD]

Do you think there is an issue with leaving the black on the sides? I've read some people capture at 720 and crop later to 640.

I'm not sure *why* they would do that. Just capture whatever the content is already, and call it good.

Modern screens are 16:9 (wide), and the tapes are probably all 4:3 (more square-ish), so there's going to be black bars anyway, whether baked into your recordings or left over from fitting it to the screen without distorting.

I would not bake them into the recording if I had a choice, because of what happens in later editing steps (possibly years or decades from now) that are not thought through. Black bars on the sides, and then top and bottom because of another shape change, and then...... After a few of these, the picture is getting pretty small with a big black frame around it...

Someone else might do that anyway, but don't start it yourself. :-)

I've included my current settings, but not much has changed. I changed the bit rate and a few of the options after gathering an average from articles I've read. Please feel free to comment on any of that if you would have suggestions.

A couple of things:

1. You're capturing directly to MP4.
   • If that crashes, you'll likely lose the entire recording, including what came before the crash.
   • If you capture to MKV, then you keep whatever came before the crash. Could be much less to redo that way.
     • A lot of things take MKV directly, so see if that works. But if you must have MP4, then OBS can convert it after the fact.
       File -> Remux...
     • You can make it automatic too, in Settings -> Advanced, so it stays as one step while keeping the crash-resilience.
       • (if it does crash, you'll have two files to splice together, after probably a manual remux, but you still don't have to re-record the entire tape!)
     • The catch to remuxing is that you now have two files with bit-for-bit identical content except for the headers, and so it takes twice the storage space. But once it's done, you can delete one and not lose anything. It's the exact same data and thus the same quality.
2. You're using a constant bitrate.
   • That's good for streaming, where you have a hard limit and you want the best possible within that limit. So you set a constant bitrate just under what you can sustain (which may be far less than the advertised or measured speed), and then tweak the efficiency settings from there while watching your system load.
     • It effectively plays with the quality setting all throughout the stream, to throw away just the right amount of detail so it can use exactly that many bits to describe what's left. Static parts of the image will become sharp, while fast and detailed motion will be blurry. So it's not so good for recording.
   • For recording, you want constant quality. Same amount of detail regardless of what's happening, and use however many bits it needs to describe that. You won't know the exact file size until you stop it and look - just runtime is not enough information anymore - but you can get a rough feel for it after a while and learn to tweak it accordingly, along with the efficiency settings that you want to have as high as they can go without bogging down the system.
     • (more work to find the best allocation of bits, and thus a smaller file size for the quality that you told it)
     • This might be useful: https://trac.ffmpeg.org/wiki/Encode/H.264

Also, as a reminder of my setup, although I am capturing at 720, I am still running into the HDMI upscaler, then out HDMI to the capture card. So I am upscaling to 1080 and capturing 720x480, which I may change.

Upscaling from 480 to 1080 and back down to 480 again. You don't get back what it was originally. Depending on how good *both* scalers are, it may or may not be indistinguishable, but it's almost certainly not the same thing.

Just capture it raw, with no scaling whatsoever. And yes, that'll probably require a different capture device, which you're already working on:

I am considering getting a used BlackMagic Intensity Shuttle so that I can use the S-video and also because this generally seemed to be rated as a great USB capture card.

Well-known and well-respected brand name. But I had some trouble with the one product that I ever bought from them: a 4x HD video capture card in a desktop PCIe slot. Even support couldn't figure it out, so I ended up sending it back for a refund.

That said though, they DID have support, that stuck with me throughout the entire process and never gave up. They even offered for me to send it to their shop for their own teardown for free. But Amazon's return window was running short by then, and I noticed that a brute-force mathematical Combinatorics exercise on all of their available settings, still wouldn't give me what I needed even if we did get the other bugs ironed out.

I also learned during that process, that this product in particular was not an original design, but a substantial modification of an older one, and so it probably hadn't been completely thought through yet. Somewhat like editing a draft a draft and not catching all the mistakes before posting it. ;-)

Not quite what I would expect on average from a well-regarded brand name, but everyone does do that from time to time, so I won't beat on them *too* bad for it.

 

[zibon]

I'm not sure *why* they would do that. Just capture whatever the content is already, and call it good.

Modern screens are 16:9 (wide), and the tapes are probably all 4:3 (more square-ish), so there's going to be black bars anyway, whether baked into your recordings or left over from fitting it to the screen without distorting.

I would not bake them into the recording if I had a choice, because of what happens in later editing steps (possibly years or decades from now) that are not thought through. Black bars on the sides, and then top and bottom because of another shape change, and then...... After a few of these, the picture is getting pretty small with a big black frame around it...

Someone else might do that anyway, but don't start it yourself. :-)


A couple of things:

1. You're capturing directly to MP4.
   • If that crashes, you'll likely lose the entire recording, including what came before the crash.
   • If you capture to MKV, then you keep whatever came before the crash. Could be much less to redo that way.
     • A lot of things take MKV directly, so see if that works. But if you must have MP4, then OBS can convert it after the fact.
       File -> Remux...
     • You can make it automatic too, in Settings -> Advanced, so it stays as one step while keeping the crash-resilience.
       • (if it does crash, you'll have two files to splice together, after probably a manual remux, but you still don't have to re-record the entire tape!)
     • The catch to remuxing is that you now have two files with bit-for-bit identical content except for the headers, and so it takes twice the storage space. But once it's done, you can delete one and not lose anything. It's the exact same data and thus the same quality.
2. You're using a constant bitrate.
   • That's good for streaming, where you have a hard limit and you want the best possible within that limit. So you set a constant bitrate just under what you can sustain (which may be far less than the advertised or measured speed), and then tweak the efficiency settings from there while watching your system load.
     • It effectively plays with the quality setting all throughout the stream, to throw away just the right amount of detail so it can use exactly that many bits to describe what's left. Static parts of the image will become sharp, while fast and detailed motion will be blurry. So it's not so good for recording.
   • For recording, you want constant quality. Same amount of detail regardless of what's happening, and use however many bits it needs to describe that. You won't know the exact file size until you stop it and look - just runtime is not enough information anymore - but you can get a rough feel for it after a while and learn to tweak it accordingly, along with the efficiency settings that you want to have as high as they can go without bogging down the system.
     • (more work to find the best allocation of bits, and thus a smaller file size for the quality that you told it)
     • This might be useful: https://trac.ffmpeg.org/wiki/Encode/H.264

Upscaling from 480 to 1080 and back down to 480 again. You don't get back what it was originally. Depending on how good *both* scalers are, it may or may not be indistinguishable, but it's almost certainly not the same thing.

Just capture it raw, with no scaling whatsoever. And yes, that'll probably require a different capture device, which you're already working on:

Well-known and well-respected brand name. But I had some trouble with the one product that I ever bought from them: a 4x HD video capture card in a desktop PCIe slot. Even support couldn't figure it out, so I ended up sending it back for a refund.

That said though, they DID have support, that stuck with me throughout the entire process and never gave up. They even offered for me to send it to their shop for their own teardown for free. But Amazon's return window was running short by then, and I noticed that a brute-force mathematical Combinatorics exercise on all of their available settings, still wouldn't give me what I needed even if we did get the other bugs ironed out.

I also learned during that process, that this product in particular was not an original design, but a substantial modification of an older one, and so it probably hadn't been completely thought through yet. Somewhat like editing a draft a draft and not catching all the mistakes before posting it. ;-)

Not quite what I would expect on average from a well-regarded brand name, but everyone does do that from time to time, so I won't beat on them *too* bad for it.

First, thank you for all of the info. Extremely helpful. I was aware of the mp4 crash issue, but not aware that I could automate the remux, so thank you for that. I will use mkv.

And thanks for that bit rate link….really informative. I will utilize that info as well.

A couple questions…should I leave my frame rate at 59.94? I did this based on various articles discussing capturing the fields.

Also, yes the tapes are 4:3. The 720x480 is suggested in a lot of forums/articles (with many also saying 640x480…and then also several other resolutions). I assume this means capturing at 720 but cropping it to 640 so it is 4:3 (eliminating the black bars)? Believe it or not from my questions, I’ve been an A/V engineer since the 90s. We always did SD work at 640x480, so this is what I thought I would use, but I kept seeing the 720 mentioned.

Again, your reply was very helpful. Thanks again!

 

[AaronD]

A couple questions…should I leave my frame rate at 59.94? I did this based on various articles discussing capturing the fields.

The raw source is interlaced, which creates some confusion today about how to count a frame in the first place:

If you must refresh EVERY line before counting a frame, that's 29.97.
If you only make one trip down the screen, that's 59.94.

Given that, the 2x de-interlacers produce 59.94 full frames per second, and the non-2x ones produce 29.97. Set OBS accordingly.

Also, yes the tapes are 4:3. The 720x480 is suggested in a lot of forums/articles (with many also saying 640x480…and then also several other resolutions). I assume this means capturing at 720 but cropping it to 640 so it is 4:3 (eliminating the black bars)? Believe it or not from my questions, I’ve been an A/V engineer since the 90s. We always did SD work at 640x480, so this is what I thought I would use, but I kept seeing the 720 mentioned.

I don't know why people use 720. I'd disregard it. Record *exactly* what the capture card gives you, except possibly for de-interlacing, since most modern things would much rather have progressive. So use a good de-interlacer, set it well (if it *has* settings), and record what that gives you, as-is.

 

[zibon]

Thank you once again for the reply.

The raw source is interlaced, which creates some confusion today about how to count a frame in the first place:

If you must refresh EVERY line before counting a frame, that's 29.97.
If you only make one trip down the screen, that's 59.94.

Given that, the 2x de-interlacers produce 59.94 full frames per second, and the non-2x ones produce 29.97. Set OBS accordingly.

When you say de-interlacers, are you referring to the hardware? Again, I am using a composite upscaler to HDMI into an HDMI capture card. So could one of these be de-interlacing? If so, is there a way to determine this so I could set OBS correctly? Apologies if I am misunderstanding.

Also, it appears I am going to stick with this setup. The BlackMagic Intensity Shuttle doesn't appear to be compatible with M1 Macs. The only other option I can think of (that is easy) are the many s-video capture devices (like the many on Amazon) but they all sound like they are not very good according to A/V forums. I've been doing a bunch of tests (creating files and casting them to my TV) and 640x480 59.94 Yadifx2 looks good to me on my TV...that's with the current settings, I haven't adjusted bitrates or changed to MKV since your last suggestions, but I plan to do that. In other words, even though this composite upscaling to HDMI into the HDMI capture device seems counter-productive, it looks good and smooth to me on a TV.

Thanks again for you help.

 

[AaronD]

When you say de-interlacers, are you referring to the hardware? Again, I am using a composite upscaler to HDMI into an HDMI capture card. So could one of these be de-interlacing?

Possibly, but don't use the upscaler to HDMI, or an HDMI capture. Get a direct VCR to USB capture. Details above.

And even so, the deinterlacer could still be in the capture card, or you may need one in OBS. Right-click the source -> Deinterlacing -> Yadif. Play with 2x or not, with the corresponding framerate in OBS, and Top or Bottom Field First, to see what gives you the best result.

The only other option I can think of (that is easy) are the many s-video capture devices (like the many on Amazon) but they all sound like they are not very good according to A/V forums.

You're not sorting by price, are you? Or keeping the "Featured" sorting that is a combination of marketing and raw popularity? (which is itself influenced by people sorting by price)

Look on the general internet, find something good there, and then restrict your shopping to just that. It might still be from Amazon or wherever, but through a constrained search that doesn't allow it to give you so much crap.

Also try to read between the lines with those reviews. A lot of the bad ones (and forum complaints) are from people who didn't have a clue what they're doing or what they're talking about, and blame the product for essentially not reading their minds to correct their mangled terminology or gross misunderstanding. That's not to say that the products are necessarily good either - all of the above still applies too - but you do kinda have to know what you're talking about yourself and filter everything.

I've been doing a bunch of tests (creating files and casting them to my TV) and 640x480 59.94 Yadifx2 looks good to me on my TV...In other words, even though this composite upscaling to HDMI into the HDMI capture device seems counter-productive, it looks good and smooth to me on a TV.

Okay, I guess? I'd still see if I could do it "right", with as few conversions as absolutely possible.