The GV-USB2 capture card and OBS for VHS and the settings

[BobbyL]

Video on Youtube https://www.youtube.com/watch?v=yPJR8u3RgFg&t=526s

speaks of setting up GV-USB2 in OBS as a 4:3 capture (1440x1080) as that is native for VHS.
Stated if you capture at 16:9 it makes images wide.

A commenter posted:
Instead of stretching the video to 1440x1080 you can set the screen to the same resolution of the capture device by right-clicking on the video output and selecting "Resize output (source size)

What is happening with the first choice vs the second choice? Is the first instruction a valid way to save the recording?

When I create a 4:3 vid and save in Pinnacle as "Same as Timeline" I get a very square image.
If I save as what Pinnacle calls HD 1080p/MPEG-4 Visual/1920x1080 I get a more pleasing rectangular viewing area with no image distortion or amplification of defects.

I saved all these videos with Canopus capture card previously - doing one last pass at the VHS before they go to landfill

Bob

[davexnet]

VHS should be captured in its native standard definition, 720x576 (PAL 25fps) or 720x480 (NTSC 29.97 fps).
Capture using a lossless codec such as Lagarith or UT Codec

[lollo]

VHS should be captured in its native standard definition, 720x576 (PAL 25fps) or 720x480 (NTSC 29.97 fps).
Capture using a lossless codec such as Lagarith or UT Codec

and use AmarecTV, not OBS

[BobbyL]

Thanks
So 4:3 ratio is not the bottom line
4/3 is 1.33 and above is 1.25 and 1.5 respectively,
don't know why "4:3" was emphasized and he took it to 1440x1080. What does that do to the image vs proper capture?

I use the Amarec TV software and add the Lagarith codec to the software? Sorry, if that is not it what are the steps?

Bob

[davexnet]

I'm more familiar with UT codec, so install it, and in AmarecTV it should show up.
Set it up like this:
https://forum.videohelp.com/threads/412235-Codec-suggestions?highlight=uly2#post2711442

[Alwyn]

So 4:3 ratio is not the bottom line

No, 4:3 is the bottom line. All standard definition video from a VCR is designed to be displayed at 4:3. Confusingly, the frame sizes for PAL are 720x576 and for NTSC 720x480, which, as you've noted, don't divide to 4:3. The 4:3 display aspect ratio is "forced" onto the video after capture when you're processing it into a viewing format such as MP4.

So the "good/best" way is to capture, as Dave says, to 720x576 (at 25 frames per second) or 720x480 (at 29.97 frames per second). Then during post-processing, deinterlace it and resize it.

1440x1080 is a valid 4:3 frame size using square pixels (the others, 576 and 480, display the pixels as non-square) and is suggested by some because the scaling onto a UHD TV is a simple double of the height (1080 doubled=2160).

For the general AmarecTV setup, see my guide here.

OBS is an alternative if you're happy going straight to MP4. It's a simpler process but not quite up to the standard of a true lossless capture and processing.

Another good OBS guide is by Tim Ford on YT.

[lollo]

If staying in SD contest, there is no need to (lossy) resize anything. In post-processing keep the original capture proportions and specify a 4:3 DAR.

If for whatever reason you upscale to 1440x1080 the pixels are “square” (width x height = DAR)

[BobbyL]

Thank you for all of that. I am planning on software above, no further OBS for now.
>>So the "good/best" way is to capture, as Dave says, to 720x576 (at 25 frames per second) or 720x480 (at 29.97 frames per second). Then during post-processing, deinterlace it and resize it.<<
So I can resize with Pinnacle, et al, as long as I capture it properly to the above? (sorry to ask)
I thought the 1440x1080 4:3 was too narrow. These are captured with OBS
I do see 1920x108o as pixel standout.

[Alwyn]

I thought the 1440x1080 4:3 was too narrow.

Yes. The 16-9 file is closer to real but shouldn't have the black side bars. The active picture area is 1.31:1, close to 4:3 which is 1.333:1. I suspect the OBS settings for the resolutions were not right. The captured video is 720x576, but the output should be set to 768x576 to stretch it to the correct 4:3 frame size.


[Attachment 81442 - Click to enlarge]

You'll find AmaRecTV will also capture at 720x576. That will need to be either stretched to 768x576 or the 4:3 flag applied so it displays at 4:3 (768x576).

So I can resize with Pinnacle, et al, as long as I capture it properly to the above? (sorry to ask)

You should be able to. No need to be sorry! That's what the forum's for.

[BobbyL]

Thanks!
I set up AmarecTV but the lossless avi file size and the lack of deinterlacing in that app (needing another app to deinterlace coz the lines are very obvious) makes OBS and the guidance here and on Tim Ford YT, makes it my choice.

I appreciate the guidance, and being shown AmarecTV, and the tweaks for OBS.
Bob

[Alwyn]

[BobbyL]

And one would think that was the end of the thread. My decision to save in mp4 format with OBS leads me to wonder (and really explode with questions)

1. Is the lossless avi of AmarecTV after deinterlacing superior to mp4 to all eyes? Or more to a professional's eyes? Average family member looking at the subjects in the video more than the quality of what they are looking at.
2. If my monitor is currently an old Toshiba LCD 720p HDTV...and it is played on whatever upgrade TV is in the family room ... are the errors in quality amplified?
This monitor review: The Toshiba 32C100U is a 720p HDTV, which means it can't natively display the highest HD formats available (10080i and 1080p)
3. Can I deinterlace with another app suggestion besides AviSynth to make these lines go away?
4. If I edit with Pinnacle (or really what editor should I use?) do I save it as avi to stay lossless?

Bob (who probably should have taken Alwyn's thumbs up and left it at that)

[lollo]

Post a sample of an OBS on-the-fly deinterlaced capture, and a lossless 4:2:2 interlaced AmarecTV capture of the same video segment. We’ll show you the differences.

[BobbyL]

Here is OBS with settings that were suggested on this forum (deinterlaced)
vs AmarecTV with suggested settings
vs AmarectTV modified with Handbrake BOB setting but not knowing what the heck I was doing with the other settings there.

The interlaced lines of AmarecTV makes it very hard for me to see the inherent superiority of that capture.

Is there a Avisynth guide for dummies?

Bob

[davexnet]

The AmarecTV avi should be a lossless capture, ie. the same data coming out of the capture device.
If you like the smoother motion "look", you can double-rate deinterlace as I did here, also some chroma noise reduction.

[davexnet]

oops... dup post!

[Alwyn]

The interlaced lines of AmarecTV makes it very hard for me to see the inherent superiority of that capture.

You deinterlace to remove the jaggies:

-In VDub, use the Deinterlace filter set to Yadif and Double Frame Rate-Top Field First

-Use AVISynth with the QTGMC filter.

In your scenario, I can't see the point in using Handbrake (which looks to me to be the worst of the lot) because for the same effort, you can use Virtual Dub to deinterlace and also easily do all the rest: crop, resize, apply the CCD filter and a bit of basic noise reduction then export to MP4.

In this case, the winner, as one would expect, is the QTGMC version. It's very hard to pick between the VDub version and the OBS version.

The OBS version certainly isn't the catastrophic/molested/garbage/vomit which some would describe it as.

For these two, I have applied the Camcorder Denoise filter in VDub to both but otherwise applied no denoising (noting that QTGMC does some very light denoising automatically). QTGMC is an AVISynth script which is opened in VDub for exporting to MP4.

[BobbyL]

OK - so VirtualDub did make things look better and now I need to understand how hard it is to insert QTGMC into VirtualDub, unless it is preloaded.
I'll figure that out coz it really helped.

And you guys really helped! Thanks

Dave, did you create your version with one of Alwyn's processes? Which?

Bob

[Alwyn]

so VirtualDub did make things look better and now I need to understand how hard it is to insert QTGMC into VirtualDub, unless it is preloaded.

Just to amplify, QTGMC is an AVISynth de-interlacing function. What you do is create an AVISynth script like this one I used:

AviSource("H:\Videohelp\BobbyL\Bobby amarec(20240818-1450).avi")
  
    ConvertToYV12(interlaced=true)
    AssumeTFF()
    QTGMC(preset="Fast", EdiThreads=8)
    Prefetch(24)

That text is saved as an "AVS" file eg BobbyL.avs and then opened in VDub. In this case, VDub is just the viewer of the AVS file (which is processing the video using the commands in that AVS/script file).

So no, VDub doesn't have QTGMC "preloaded".

I'm in the early stages of a "Simple AVISynth" guide which will help people set it up easily. Standby for that...

[davexnet]

Dave, did you create your version with one of Alwyn's processes? Which?

Bob

Similar, I created my own script and encoded in vdub2

import("I:\plugins\AviSynth+021224\plugins\mtmodes_062121c.avs") # load mtmodes for prefetch
aud=directshowsource("C:\Users\public\Documents\amarec(20240818-1450).avi",video=no)  # load audio
vid=lwlibavvideosource("C:\Users\public\Documents\amarec(20240818-1450).avi")      # load video
audiodub(vid,aud)                                             # merge audio and video
assumetff()        # top field first
qtgmc(tr2=1)     #    double rate deinterlace with some cleanup
convertToyv12()  # convert color space to consumer safe yv12
cnr2()                # chroma noise reduction
prefetch(2)         #  two threads

[lollo]

The interlaced lines of AmarecTV makes it very hard for me to see the inherent superiority of that capture.

Your OBS capture is missing half of the motion and is less detailed; it is also resized to 768x576 with a wrong Display Aspect Ratio flag. I suspect is also missing some frames, but did not check in details.




For the rest, get familiar with interlaced video concepts, capture lossless YUV 4:2:2 interlaced with AmarectTV and deinterlace with QTGMC in AviSynth.

[Alwyn]

Your OBS capture is missing half of the motion and is less detailed; it is also resized to 768x576 with a wrong Display Aspect Ratio flag.

How about comparing apples and apples. Encode the AVI at 3500kbps and then compare them. 768 is obviously simply an incorrect OBS setup and the DAR is 4:3, which is correct.

[lollo]

How about comparing apples and apples. Encode the AVI at 3500kbps and then compare them.

OBS capture is missing half of the motion and is less detailed. Whatever encoding is behind.

768 is obviously simply an incorrect OBS setup and the DAR is 4:3, which is correct.

768x576 implies DAR=1, whatever error is behind.

[Alwyn]

768x576 implies DAR=1, whatever error is behind.

So DAR of 16:9 for 1920x1080 files is wrong? 1080x1080 can have a DAR of 4:3. Just like 720x576 has a DAR of 4:3. They all display in a 4:3 frame. Unless the width/height=DAR, the pixels will be stretched or squeezed. 768x576 are square because no squeezing.

[Sharc]

Here apples with apples in slow motion. The smooth and sharper amarec vs the jittery (missing half of the motion) and blurred baserate OBS. Make your choice.

[Alwyn]

Sharc, is the OBS file you used in the stack the original from Bobby because if it is, you're being unrealistic. It was low-bitrate anyway (which can be easily increased in OBS) , it's been slowed down, which is not how people view it, and the original OBS file looks nothing like that jumpy mess you've created.

[BobbyL]

Wow. Such dissection of the OBS and all true!
I thought I had followed Tim-From-YT video but bitrate was set at 2500, here is 3500.
I had 720x540 set but thought Alwyn was suggesting 720x576 in post #9. If not, then I totally missed his point. Changed to 720x540.
BTW - Tim said to set canvas at 720x480. Is that correct?

I likely will have to stop dead here if I can't find a step by step process to set up QTGMC in VirtualDub, let alone in Avisynth.
I appreciate the scripts, assume it is done in Notepad, but don't know how to modify the script for my machine and folders, where to place in VirtualDub, let alone Avisynth, and my mouth was agape trying to follow this dude, who probably is making instructional videos for an n=5.
https://www.youtube.com/watch?v=JUWVt1Q7Afo

[BobbyL]

BTW - my family and future generations of family will likely never care about quality, or maybe never even watch these (hoping for one historian to emerge in the future).
But now I see how beautiful AmarecTV deinterlaced is! And I would like to learn to do that. Thank you for showing me that and Sharc for slowing it down. Like VHS butter!

[Alwyn]

I had 720x540 set but thought Alwyn was suggesting 720x576 in post #9.

If it's PAL, yes, the capture is set to 720x576 but the output is to 768x576. For NTSC, 720x480 and 720x540.

BTW - Tim said to set canvas at 720x480. Is that correct?

If it's NTSC, yes, correct.

But now I see how beautiful AmarecTV deinterlaced is! And I would like to learn to do that. Thank you for showing me that and Sharc for slowing it down. Like VHS butter!

Sharc has to explain what he's done here. Your original OBS file in post #14 isn't jerky like that at all at normal speed. And you will not watch these at half speed. Note my comments in post 26. He should also acknowledge that the source OBS file is only 29.97fps, whereas I'll bet the AVI he's used for his test is deinterlaced to 59.94. Of course it's going to be smoother.

OBS can easily be set to produce a 59.94fps (ie deinterlaced) MP4 (for NTSC).

Watch each video (with the AVI rendered to MP4) and you will see the difference is much less. I can tell the difference, but it is entirely up to you, on behalf of the viewers, to create something that passes the cost-benefit test. Would I choke if I saw your OBS one in isolation? Probably not. Will your family, most probably not.

BTW - my family and future generations of family will likely never care about quality, or maybe never even watch these (hoping for one historian to emerge in the future).

That's the point. Is it worth the extra effort? In my case, yes, I've got a fast computer that processes QTGMC quickly. I've spent hundreds of hours learning the processes as a hobby, and I haven't even touched the surface of what the experts here know. If a video's going to be watched once or twice, is it worth the extra time and effort to you?

[Sharc]

Sharc has to explain what he's done here. Your original OBS file in post #14 isn't jerky like that at all at normal speed.

Comparing at same timeline and same compression (apples with apples, ~8...9MiB .mp4 files). Making the gap of temporal and spatial resolution a bit more eye-catching. The OP has understood it.

OBS can easily be set to produce a 59.94fps (ie deinterlaced) MP4 (for NTSC).

Please do it. I can only compare what we got. Being an OBS ignorant I have several times been asking here for a decent sample of an interlaced OBS VHS capture from OBS experts, but I didn't receive a reply or just got some useless and flawed crap. I am looking forward to eventually getting one, really. So please help me, and I will be happy to compare again and stand corrected

Watch each video (with the AVI rendered to MP4) and you will see the difference is much less. I can tell the difference, but it is entirely up to you, on behalf of the viewers, to create something that passes the cost-benefit test. Would I choke if I saw your OBS one in isolation? Probably not. Will your family, most probably not.

Agree. If on a 5 star quality rating one is happy with say 2 stars instead of getting say 4 stars with about same efforts by just doing it right, I have nothing to add, especially when there is nothing to compare with.
Or when one prefers a shiny GUI with endless setting options over decent video quality I have nothing to add either. But I admit that I am an OBS ignorant still looking forward to eventually getting a decent interlaced VHS capture out of it, comparable to AmarecTV or Vdub. Your help or an example for a better 'apples to apples' comparison would be appreciated.