Advice needed for converting family VHS tapes to digital

[jagabo]

Here's a sample.

*** Edit: the sample I took (even though I set it to 59.94) still shows in MediaInfo as 29.97.

What about the first video in post #86? MediaInfo says it's 59.9f fps.

[jadoggin]

I must have overwrote that. But I don't want to waste any more of my life with VirtualDub so I'm going to just use EZ Grabber and be done with it. I guess I can still use VD for the histogram, although I'm not sure what I'm really looking for. Can you tell me how to read it? Here is an image I took just now of the histogram when the video was playing.

[jagabo]

VirtualDub's histogram is a true histogram (as opposed to Histogram() in AviSynth which defaults to a wave form monitor). At the far left is intensity (Y) 0, at the far right is intensity 255. In between are the other 254 intensities. The height of the graph indicates how many pixels had that intensity. Video should be kept predominantly between 16 and 235. The "illegal" areas (Y<16, Y>235) are marked in red, the legal areas (16-235) are marked in blue. A little red is usually ok because it's usually from oversharpening artifacts, represents a very small portion of the picture, or the black borders at the left and right of the frame (which you don't really care about).

[jadoggin]

So what the histogram shows there is perfectly acceptable. Got it. Ok, one more question. I just plugged in the Panasonic DMR-ES10 to be used as the TBC. So I am running only video through that right? Current configuration is Left and Right audio go straight from VCR to USB capture. Then I have the composite video going from the VCR to the ES10, then an S-Video cable from the ES10 to the capture device. Is this the correct configuration, or should the audio cables also be passing through the ES10?

[jadoggin]

So I took two different captures - one with just the VCR (first screenshot), and one with the ES10 as the passthrough TBC (second screenshot). Why does the cap with the TBC look like it has so many more artifacts? I know the screenshots I uploaded are not the exact same frame, but in my opinion, the capture with just the VCR looks better. Is that normal?

[smrpix]

Why does the cap with the TBC look like it has so many more artifacts?

Because it has sharpened as well as stabilized. The best way to judge which you prefer, however, is in motion -- not stills.

[jadoggin]

Well I can certainly tell it is more sharp - which I like, but it's also quite a bit more noisy as well. Is there a way I can clean that up?

[LMotlow]

I don't think anyone here will advise how to clean still photos or screen shots of media players. Better to have a sample of real video. A few seconds of original, unprocessed capture is what's usually submitted. If you need instructions for that, just ask.

[jagabo]

And the sample images are poorly deinterlaced.

Go through the DVD recorders settings and turn off any sharpening and noise reduction filters.

[jagabo]

So what the histogram shows there is perfectly acceptable.

Maybe. There are some pixels below Y=16. Make sure there aren't any dark shots with lots of pixels down there.

Your sample video in post #89 is showing blown out brights and some too dark blacks:



The black bar at the right of the frame is down around Y=0. It's normally at Y=16. You don't particularly care about that bar being too black but the implication is that there could be other parts of the recording where the picture is down there too. There are some pixels below Y=16 in the main part of the picture but I think those are mostly over sharpening halos.

Ok, one more question. I just plugged in the Panasonic DMR-ES10 to be used as the TBC. So I am running only video through that right? Current configuration is Left and Right audio go straight from VCR to USB capture. Then I have the composite video going from the VCR to the ES10, then an S-Video cable from the ES10 to the capture device. Is this the correct configuration, or should the audio cables also be passing through the ES10?

It probably doesn't matter. The DVD recorder's processing probably causes a slight delay in the video. I don't know if it delays the audio to match. You might test for that.

[jadoggin]

I don't think anyone here will advise how to clean still photos or screen shots of media players. Better to have a sample of real video. A few seconds of original, unprocessed capture is what's usually submitted. If you need instructions for that, just ask.

Here is a sample of my latest capture using the TBC with EZ Grabber.

[jagabo]

The time base jitter has been replaced with occasional shifts left/right of part of the frame. The flagging at the top of the frame is improved though. What's with all the little dropouts? I didn't see dropouts like that in the VirtualDub caps. Attached the original (only an IVTC) on the left, slightly cleaned on the right.

If the dropouts are different each time you capture you can capture three times and then use a median filter to get rid of them.

[newpball]

Pretty bad, the shifting, dropouts, a lot of bad noise.

What kind of VCR player are you using?

[jagabo]

Capture the same segment without the TBC and compare.

[LMotlow]

And make sure your VCR is plugged into the Line 1 input of the ES10. Depending on the production run, some ES10's don't have line tbc on the other inputs. Your sample is also Bottom Field First. Did you capture to DV-AVI? Tape almost always plays as Top Field First (TFF), while DV-AVI is always BFF.

This video can't be deinterlaced, in case you might want to do that later. It's hard-telecined (3:2 pulldown).

[jagabo]

The capture device determines the field order of the capture frames. It doesn't matter if you capture top field first or bottom field first. With film sources either will inverse telecine properly. True interlaced video is neither tff, nor bff, it's just an alternating sequence of top and bottom fields. So it doesn't matter whether you capture it TFF or BFF. You just need to treat the video properly downstream.

[LMotlow]

I don't agree with that at all, as TFF playback capped as BFF always has subtle peculiarities in every TFF->BFF capture I've ever dealt with and is always a bitch to smooth over. But let's not digress. The first task is to get a decent capture, whatever the field order.

[Cornucopia]

1, 2 3, 4 5, 6 7
is the same as
1 2, 3 4, 5 6, 7

If they're all different at different time periods, the only difference is how one groups them together.

Now, if they get put out of order, that's a different matter altogether!

Scott

[LMotlow]

Yep. I know how it's suppose to work. Supposed to. Theoretically. Doesn't matter here, the O.P. won't be getting TFF anyway and wouldn't notice the difference.

[jadoggin]

Ok, so I moved my captures to my laptop. I just don't think my desktop has the juice for this. I still get the split frame when trying to capture in VirtualDub. I'm still troubleshooting why. For now EZ Grabber is working, but I'd like to know what to use to clean up the noise. NeatVideo? Here are two captures from EZ Grabber - the first (Capture1) is straight from the VCR, the second (Capture2) is with the ES10 and VCR. Line into the ES10 has always been in Input 1. I didn't realize this project would require so much patience .

[2Bdecided]

Is it all Veggie Tales? Some of the DVD boxed sets are pretty cheap, and many are on Amazon Prime Instant Video.

For some kids shows I might make an exception, but generally if there's another way of getting something in decent quality, I won't bother transferring it from VHS.


The ES10 TBC is working OK. It's also sharpening the image, which looks OK here to my eyes, but I would want to know how to turn it off for other tapes. Unfortunately it's increasing (or maybe even generating) chroma noise to an unacceptable level. Worse of all, on both caps, there's horrible chroma noise in the headlights. Something is very wrong there, either with the VCR or capture device.

Cheers,
David.

[jadoggin]

No, the Veggie Tales tape I have is only for testing. I don't want to test on my 25-year old actual tape until I have method to this madness. The Veggie Tales tape seemed like a good way to test. Anyway, I started with an Osprey capture card and the VCR only and so far the capture I took (using Power Director) with that combination looks the best to my eyes. But, it was recommended that I ditch both that card and software to upgrade to this USB capture device while using the ES10 as a passthrough. Now it appears I've gone backwards .

Are you saying the chroma noise is the same on both caps or that it's worse with the ES10? So if I capture from the ES10 while playing a DVD and using the USB capture, you should be able to tell that it's either the VCR or capture device, right? I'll find a DVD and grab a quick capture and upload it here.

[LMotlow]

Unfortunately it's increasing (or maybe even generating) chroma noise to an unacceptable level. Worse of all, on both caps, there's horrible chroma noise in the headlights. Something is very wrong there, either with the VCR or capture device.

I agree, but don't know where the noise is coming from. The noise in the brights is on both captures, worse in the cap without the ES10.

Capture1 Frame 270 of the original without the ES10 (not to mention dot crawl):


The player is oversharpening film grain. Likely the player needs alignment -- without a tbc, the frame rate is 30.7fps. It's exactly 29.97 with the ES10. Player has a problem with blue wavy noise and streaks in reds. Maybe it's the capture card or the tape or the player, hard to say, but the blue streaky junk appears in both captures, more distinct with the ES10.

After inverse telecine and a quick look:

Capture1: it's not sharper, it's fuzzy, but some fine details hold up better than with the ES10. Dot crawl. The side border is bent at the top, with occasional border notches as the video plays.

Capture1 original, frame 71 (blue noise visible in red, but not as much):


Capture1 filtered, frame 56:


Capture2: the tbc works, but there's something haywire with your ES10. Some fine details are murky and smeared (are all noise reduction settings in any of the menus turned turned OFF? They should be).

Capture2 original, frame 74:


Capture 2 filtered, frame 59 (stronger filtering is making eyebrow contours disappear more):


NeatVideo: it won't fix the blue distortion, and it will remove grain to the point where you'll get banding and macroblocks when encoding. Film source has fine grain that requires careful filtering. Remove 100% of the grain, you'll see plenty of other noise later.

Your samples are uncompressed AVI. Later you might try lossless recompression with huffyuv or Lagarith into YUY2. You can't use the original colorspace in Avisynth, and you'll need YUY2 and/or YV12 for Avisynth filters.

You need a better player. You need a better source, if possible.

[jagabo]

You need to figure out where each of the problems originate.

I somewhat disagree about the TBC working. There are times when it improves the horizontal jitter but there are times when it makes it worse. It does straighten out the bend at the top ~half of the frame.

The DVD recorder is also increasing the high frequency noise in both the luma and chroma channels. Some of that is probably caused by a sharpening filter. If you have control over that -- turn it off. Blurring reduces high frequency noise, sharpening increases it.

The sparkling dots in the headlight of the first shot are in both captures. I suspect the root cause is the same for the sparkling dots in the rest of the video, it just shows up more in the bright lights. They are too sharp (especially in the chroma channels) to be "on the tape" so it's happening somewhere in the VCR, or capture card.

Without TBC, Y, U, and V as greyscale:


With TBC:


LOL, I see that LMotlow posted the same YUV images. In case you're wondering, they were made in AviSynth with StackHorizontal(GreyScale(), UtoY(), VtoY()).

[LMotlow]

Interesting notes , jagabo. My ES15 or RF-XS34 as tbc don't do any of that -- but I paid plenty for both. Can't say I've seen any of it before. I don't think the ES10 has a sharpener??? But my old ES10 went RIP some years ago, I don't recall the menus.

[jagabo]

Interesting notes , jagabo. My ES15 or RF-XS34 as tbc don't do any of that

My ES15 doesn't either. And nobody who posted in the "Who uses a DVD recorder as a line tbc" thread had such problems:

https://forum.videohelp.com/threads/319420-Who-uses-a-DVD-recorder-as-a-line-TBC-and-what-do-you-use

[jadoggin]

So hopefully this will clear up some of the confusion. I grabbed a capture straight from the ES10 using a DVD as the source. See attached video. From what I can tell, the entire issue is the VCR, or possible the VCR tape. I'm going to try a different tape to see if the problem exists with that as well.

[LMotlow]

Hold on. Are you saying that the earlier captures were from a tape that was made from a DVD? Well....you don't record a DVD or capture it. DVD is copied directly to a computer using VOB2MPG , or a decrypter that does the same thing if the DVD is commercial stuff.

Capture 3 looks better. Kinda soft but you can fix that. Will have a closer look. No sound (the capture has an empty audio track). The frame rate is 33.139fps(!)

[jadoggin]

Hold on. Are you saying that the earlier captures were from a tape that was made from a DVD? Well....you don't record a DVD or capture it. DVD is copied directly to a computer using VOB2MPG , or a decrypter that does the same thing if the DVD is commercial stuff.

Capture 3 looks better. Kinda soft but you can fix that. Will have a closer look. No sound (the capture has an empty audio track).

No. All my previous captures were from an original VCR tape (none of the VCR tapes were made from a DVD). I'm quite familiar with copying a DVD to a computer and none of what I'm doing involves any of that. Capture3 is simply to show the quality using a DVD source directly from the ES10 to prove that the Diamond USB capture device is not the cause of the bad quality. I intentionally never captured the sound as I only wanted to show the video quality. Which I think we can safely say that the cause is either the VCR tape or the VCR itself.

I've also attached a video clip showing all the settings I have configured for my ES10.

[jagabo]

So hopefully this will clear up some of the confusion. I grabbed a capture straight from the ES10 using a DVD as the source. See attached video. From what I can tell, the entire issue is the VCR, or possible the VCR tape. I'm going to try a different tape to see if the problem exists with that as well.

That's definitely a lot cleaner. Free of the speckles in the luma and chroma channels. But I'm not absolutely sure this indicates the problem is in the VCR or the tape. It could be some interaction between the components. For example, a ground loop problem might manifest differently depending on which components are attached. Also the output levels of the DVD are more in line with standard levels. There are no blown out brights.

In addition to a different tape and different VCR, some other things you can try:

1) Don't connect the audio at all. That eliminates one possible ground loop source. You don't need audio to diagnose the current video problems. Alternately, run audio cables from the VCR to the DVD recorder and another pair from the DVD recorder to the capture device. Ie, don't bypass the DVD recorder.

2) Make sure all the devices are plugged into the same outlet or power strip.

3) Record on the DVD recorder and capture at the same time on the computer. You will then know that the DVD and the computer cap received exactly the same signal. Compare the DVD to the cap. Do both have speckles? Are they in exactly the same place? Of course, the DVD will have been MPEG encoded and reduced to YV12 so you expect it to be a little different than the computer cap.

4) Play the DVD recorded in step 3 and record on the computer. Compare the DVD to the cap. Repeat this with the VCR connected to the DVD recorder and playing the same tape.

5) Put your capture device in another slot if possible. Try to keep it as far as possible from other cards cards (if any). If you have other cards, remove them. Being closer or farther from other noise sources in the computer may make a difference.