VHS capture workflow optimisations?

[Nitro87]

Hi guys

Alright, so I'm finally up and running - got all my workflow, and now for the good stuff - omtimisation. I am a little underwhelemed by the quality but I think that is because I may not be doing something right. Initially, it was preset at 240 lines and looked horrible, even for 240 lines - VERY blocky - I corrected this to 720x576 and it looked much better - though still not 100%.

Workflow is: JVC HR-7955EK > TBC > Pinnacle 710-USB ? VirtualDub 1.9.11 (LS's download) > AVI > Adobe Premiere Pro > MP4

First Issue - TBC

I have included three sample AVI videos of the same clip: https://drive.google.com/drive/folde...fu?usp=sharing

- No Frame TBC
This is direct to the Pinnacle 710-USB with no TBC (although line TBC is on from the VCR). There are no dropped frames, and audio looks in sync (though with such short clips I know drift won't be noticeable yet) and as you can see, the image is covering the full frame bottom to top.

- AVT-8710 (Green)
Same as above, though you'll notice that this has inserted a black mask at the top and bottom of the image. This isn't what I want, I want the full image to be displayed. Is this a feature of this particular TBC? Can I change this somehow?

- Lektropacks CTB-520
This looks good but there is some slight flashing on the screen. Not sure the correct term. I had to mess with the controls to get what I think is the right colour.

I think the AVT is the best - I guess I'm asking is there a way to get rid of that black mask at the top and bottom?

Second Issue - Interlacing

I don't expect VirtualDub to deinterlace, and I think any deinterlacing should be done when I put it into Adobe anyway but what is weird is that it looks interlaced but it isn't. Adobe says it's progressive, TMPGENC says it is interlaced, and my eyes see it juddering on my TV during playback - if the TV thinks it is progressive, then it won't deinterlace.

Can I get some opinions on this please?

Third Issue - Compression
It looks a bit blocky. I know this is 576 resolution, but I expected it to look a little better. Is it just me? Is YUV2 the way to go? Are there other lossless formats I should try? Have I optimised enough?

[lordsmurf]

You need to include smaller clips attached to the thread. For example, I often answer forum threads from a tablet, and I'm not downloading 500 mb clips. Zero need for it. Sometimes images are good to show some of these issues, clips not required yet.

TBC don't insert masks. You may be seeing overscan, and/or different image offset.

Lossless files lack AR or interlacing flags. It's raw video. It must be set when used.

Adobe is terrible at deinterlace. Use QTGMC in Avisynth, convert to deinterlace before feeding to Premiere. Use Hybrid as a cheat, if needed.

Why use Premiere?

For that matter, why deinterlace? What's the reason for it?

Blocky? Are you referring to the captures, or your encoding output?

[aramkolt]

I can't speak to all the issues here, but I will point out that the AVT-8710 is adding heavy regularly spaced vertical banding, which is mostly noticeable on both of their shirts. Granted, I am viewing this on a MacBook Air using VLC, but that is not present in the other two samples. He's a cropped screen cap where you can see it. For this reason, I would say the AVT-8710 looks to be the worst of the 3, though they all probably have slightly different video levels. I do see the black bar up at the top and haven't worked with an AVT8710 to know if that is expected or not honestly wouldn't mind a TBC that blanks the head switching noise area (which is at the bottom), but maybe the top part is it getting overzealous about the upper vertical blanking area where macrovision usually is perhaps and blanking more lines than it should?

Note vertical repeating pattern in the shirts on AVT-8710:
Screenshot 2024-03-19 at 10.25.37 PM.jpg

[Nitro87]

Thanks both.

Ok, lets tackle one issue at a time. My video compression was set to (No recompression: YUY2) and is down as 1.0:1 and my CPU usage was 0% during capture. What does this mean?

I have installed HuffyUV and it is now 2.6:1 and my CPU usage is 0% but sometimes jumps up to 3%. Is that better or worse?

I really need help understanding some of the basics here before I start capturing a load of tapes.

[themaster1]

just use the old version huffyuv 2.1.1 (compression method= predict left ) yuy2

[Nitro87]

Is that better or worse than (No recompression: YUY2)?

I've just found out that the AVT-8710 using PAL will add those black bars. Happy to be corrected on this.

I tried the CBT-520, no black bars but do get some flashing imagry.

For shits and giggles, I hooked up my Sony DVD Recorder as a pass-thru. It doesn't add the black bars, it only adds some small white digital spots. This one can output both PAL and NTSC. The colour seems perfect.

Here's a screenshot... is this my best option?

[lordsmurf]

In the above attached image, those are pillarbars, and those are normal and desired. Your image is not 720x, only the palette is. The actual image is 704x. There are technical reasons that pre-date all of us.

Some TBCs do not properly retain aspect ratio (AR). Technically, there's always some % of slop in TBCs, VCRs, and capture cards. But some are egregious, unacceptable, out of bounds. too sloppy. Though none of the models mentioned here should be that way.

All TBCs, VCRs, and capture cards have different offset, image is never centered. When gear is stacked, the image can drift farther off-center depending on the gear. Remember, this is analog, not digital. There never was perfection, never centered, etc.

Sometimes people are wrong about the image size, and closer inspection reveals it.

[Nitro87]

Sorry, I'm not talking about the pillarbars. The above attached image is actually going to be my target setup as the DVD recorder passthrough looks much better than any of my TBC,s unfortunately.

I'm talking about the bars that the AVT-8710 puts on the top and bottom (horizontal) overlaying actual image. The CTB-520 doesn't do that, and neither does the Sony DVD recorder (and obviously neither does capturing on the Pinnacle directly).

The guy I bought it from says it does that on PAL signals (this is indeed a PAL tape) and it is less prominent on NTSC tapes.

[Gary34]

Quote:

I don't expect VirtualDub to deinterlace, and I think any deinterlacing should be done when I put it into Adobe anyway but what is weird is that it looks interlaced but it isn't. Adobe says it's progressive, TMPGENC says it is interlaced, and my eyes see it juddering on my TV during playback - if the TV thinks it is progressive, then it won't deinterlace.

Adobe won’t keep your colorspace either. It converts your Huffy codecs YUV colorspace to RGB when it’s imported. Adobe is a good color corrector. Do all your deinterlacing, colorspace conversion and all that in hybrid or AVSPmod. Then if you want to use Adobe afterwards.

If you are wondering things about your video information download the mediainfo app. Hit tree and you can see everything about your video.

[lordsmurf]

DVD recorders (specific models known for passthrough) do not contain frame TBCs, but merely crippled line TBCs with non-TBC frame syncs. That has consequences, and often results in duped frames. The DVD recorder output is simple clock sync'd, and not buffered timed release. It's just not a TBC.

Some people are anal about colorspace, and I'm not one of them. It's just the cost of processing, including editing. However, be aware of it (ie, what gary34 is doing here), and mitigate when possible. (The A<>D conversations can get equally ridiculous. To me, most of it is just standard internet "griping for sport" activity, and some people try to weaponize it. Worry about the major issues, not fringe aspect that have neglible effect when done properly. So colorspace, A<>D, and some others.)

[Nitro87]

I hear you.

The masking of the image, except if it's 20 pixels or so, is a dealbreaker for me hence why I think the AVT-8710 is no good for me.

I need to test some more captures with the DVD recorder as a pass through to be sure it fits all my needs.

On another note, am I using the best lossless process in VirtualDub? If so, I can move onto AvSynth/AVSPMod

[aramkolt]

Guess it depends what looks better to you. I personally can't "unsee" the repeating vertical line distortions that make it look like you're looking through a window screen or faint "jail bars" as I linked above. It's in the entire image, just easier to see on the shirts. I haven't really tested my TBCs yet, but I wouldn't consider using any that add patterns or herringbone noise to the image. The testing I plan to do is more along the lines of jittery tapes passed through a distribution amplifier and captured with multiple cards at once with different TBCs in front of them to see how they each handle jitter with an identical starting signal. With an ideal frame TBC, you shouldn't see any obvious negative effects added to the image itself as it passes through.

[lordsmurf]

The jail bars may be caused by the s-video cable. The AVT-8710 is not known for this, that is unusual to see. JVCs can have it appear in the blue screens. And then it can be an interaction issue, device crosstalk (aka, why I test workflows as workflows, not random gear tossed together). There is some art to this, it's not all science.

You're mostly testing line TBCs, and maybe capture cards? But you're in for an adventure, unless you just do flimsy superficial testing. Transparency and strength will vary wildly, especially when interacting with sources.

[hodgey]

I wasn't gonna comment on this otherwise but masking 20 pixels on the top seem out of the ordinary so wondering if this unit could have some custom/different than normal firmware? I've never seen anyone show a AVT or related TBC mask the head switch area either for that matter though I have a panasonic mixer (WJ-AVE50) that does that so at least there are other devices that do this.

[aramkolt]

Quote:

Originally Posted by lordsmurf

You're mostly testing line TBCs, and maybe capture cards? But you're in for an adventure, unless you just do flimsy superficial testing. Transparency and strength will vary wildly, especially when interacting with sources.

Mostly testing full frame TBCs as most line TBCs are built into VCRs. Though I'll probably also eventually post comparisons of some different line-TBC containing VCRs. I have a couple tapes that have a fair amount of consistent vertical jitter I'm really curious to see what they'll do. The trick will be giving each TBC an identical signal which I think can most easily be done with a single VCR playing into a distribution amplifier which will preserve the time base errors. Issue with that is you need to capture all of those streams simultaneously using different hardware and that gets logistically complicated and also adds some variables unless you have a bunch of identical capture cards (which I do not). Another way to go about it would be those small LCD video panels that can display composite inputs and videotape those all coming from the different TBC passthroughs. Any gross jitter should be pretty visible on poorly performing TBCs and it would allow focusing on the better ones in more detail via other means. Still looking for a couple more TBCs to add to the comparison before doing it though.

[Gary34]

Quote:

Another way to go about it would be those small LCD video panels that can display composite inputs and videotape those all coming from the different TBC passthroughs.

So you’re going to rely on the results that are being put out out by the ADC in an LCD device? What LCD device are you talking about?

[aramkolt]

Quote:

Originally Posted by Gary34

So you’re going to rely on the results that are being put out out by the ADC in an LCD device? What LCD device are you talking about?

Ideally what I'd want to do is store the poor composite or S-Video raw waveform signals and be able to play them back into each device individually so you start with an identical (with big timebase errors) signal. This may be possible with a DAQ, but I haven't found any recommendations on a specific device that is capable of that. CVBS_decode can capture a composite signal waveform, but I don't believe it can "play it back".

The idea with a live VHS playback plus distribution amplifier is that you need a way to then capture all of your different TBC's outputs that are working at the time time which means a lot of identical capture cards and a PC for each would be needed in the same vacinity. So if you are testing 5 TBCs at once from a single VHS playback, each TBC output has to be captured with a separate capture card and then compared later. The idea between displaying each output on small screens will give a rough idea of which cards choke, freeze, or don't resolve significant jitter. It would be be more of a screening tool to show which TBCs are worth considering in more detail for that particular type of error you're testing. I'd be useless for looking at actual picture detail, color levels, clipping etc. We'd be talking about big errors such as continuous vertical roll that would be easy to see a mile away. If the TBC can handle big errors, small ones should be no problem.

The sort of multpanel LCD screen that you could then record regular video camera would be something like this: https://www.ebay.com/itm/144986127904

[Gary34]

I hate to comment too much on the testing thing because it’s not the topic of the thread. I get off topic too sometimes. Maybe it would be a good idea to do a TBC testing thread and you can update on how everything is going or get ideas for stuff.

I can see why recommendations for different TBCs is said and not shown in video form. You would have to test across a wide variety of tapes that were shot with a bunch of different cameras. Even if you do a wide variety of tapes and even VCR combinations it’s not going to be as wide of variety as the people that have done this for 20 or 30 years. Idk maybe it depends on the VCR TBC combination too. Anyways I’m not trying to hate on the project. Good luck with your testing.

[Nitro87]

I have thoroughly enjoyed learning stuff over the last couple of days.

AVSPMod and AviSynth - I even did a script to crop a couple of pixels but still maintain aspect ratio and resize it. I learned out the scripts work, learned out to output the .avs file using FFMPEG - really seeing what you guys have been saying about "deal with stuff post capture".

I installed some plugins too.

Are there any best practice scripts for post production for VHS capture? I know every capture is different, but are there common basics that should be done - perhaps to clean up the image a bit etc?