VHS to digital - help

[newpball]

Well, if you ask what you should do with a video judged solely on one still frame of partial views of uninteresting objects, one of which has its highlights blown all to hell, in an off-color image resized to dimensions invalid for any standard video format except for PC playback -- as opposed to submitting a real piece of video that we can look at and give decent technical advice for -- but based entirely on the still image you posted, I'd say it's entirely up to you. One thing I don't get is what someone who doesn't understand basic video encoding formats and aspect ratios would be doing with Vegas Pro. But, again, that's strictly up to you.

His question is basically about cropping/masking.

It is a good question: bottom is bad, rags left and right, what do YOU do?
Seems like a straightforward question.

[freee]

One thing I don't get is what would be doing with Vegas Pro except for some cut and join work. That's a lot of money to spend just for an "editor". But, again, that's strictly up to you. With one still shot as a guide, I'd leave the video as-is.

Well since you asked, video editing is not just about encoding. That is the last part of the process and it is in the technical domain not really in the creative one (although some may argue otherwise). I am perfectly fine using Vegas Pro for video editing and I was using it even before it was acquired by Sony. What is at disposal to me in Vegas Pro is more than adequate to enjoy HD video editing with very good output by using the provided rendering templates. Believe me, I am quite good at it.

So due to my limited "encoding expertise" I resorted to asking a question here.

One still shot is more than enough to demonstrate what I am after since all frames in the video will have the same issue. Thank you for the advice given, I will leave it as it is.

[LMotlow]

I don't believe you're that good at it, or you wouldn't be asking the cropping question in the first place and I think plenty of experienced readers would agree with me on that issue. But to re-answer: seeing as PC playback is all you want anyway, crop off the bottom noise, crop off the unwanted borders, leave the video otherwise as-is, and save it to some form of DV-AVI for archive, or even to a lossless format. If you want a smaller archive file, encode as square-pixel mp4 or mkv at high h264 bitrates.

[jagabo]

Crop the borders and encode with the same pixel aspect ratio as the source, 12:11.

[usually_quiet]

No, VLC is a pretty good player.

Here is a tricky one 1:1.667 aspect ratio (square pixels!):


[Attachment 28540 - Click to enlarge]
Plays just fine in VLC!

How about if we upload it to Youtube, Youtube will re-encode but can it handle the aspect ratio?

http://www.youtube.com/watch?v=yTtD3_rAbhA&feature=youtu.be

Yes, no problem.

Only because you lucked out.

Or maybe he actually knows what he's doing and doesn't care about little plastic discs.

Indeed the poster was clear about not being interested in putting this on a DVD.

How about 1920x400?

[Attachment 28541 - Click to enlarge]

Plays fine in VLC, it even plays fine in WMP.

And youtube:

https://www.youtube.com/watch?v=qJ1jy7DSNe0&feature=youtu.be

No problems.

VLC is a good player in some ways and not so good in others. Your samples played OK with VLC. I'll give you that.

I also tried playing the samples with my 720p TV's media player and a Blu-Ray player that plays media files to see what would happen. The TV gave me a choice of "Original", which attempts to do a pixel-for-pixel mapping, and "Full" which attempts to scale the video to fit within a 1366x768 frame using its original aspect ratio. Both of those worked OK, although the 1920x400 video was treated the same in both cases because the TV's horizontal resolution is smaller than the video's actual resolution.

The Blu-Ray player's media player gave me a choice of "16:9 Full" or "16:9 Original". "16:9 Original" gave a better result. 16:9 Full attempts to fill the screen both horizontally and vertically, resulting in distorted proportions. "16:9 Original" attempts to scale the video to fit within a 16:9 frame using its original aspect ratio with pillar boxing or letter boxing added where it is needed.

Scaling the 1:1.667 aspect ratio video did produce some artifacts at 45 degrees, 135 degrees, 225 degrees and 315 degrees.

The OP's capture uses non-square pixels. It would be interesting to see what would happen with non-square pixels, but I don't have time to play with that now.

[usually_quiet]

Crop the borders and encode with the same pixel aspect ratio as the source, 12:11.

How did you arrive at a 12:11 pixel aspect ratio? The source, the OP's capture, is 720x576 (for a 5:4 SAR) with a 4:3 DAR. PAR = DAR/SAR (4/3)/(5/4) = 16:15.

[jagabo]

Crop the borders and encode with the same pixel aspect ratio as the source, 12:11.

How did you arrive at a 12:11 pixel aspect ratio? The source, the OP's capture, is 720x576 (for a 5:4 SAR) with a 4:3 DAR. PAR = DAR/SAR (4/3)/(5/4) = 16:15.

It's DV so the 4:3 image is in a ~704x576 portion of the frame.

[Skiller]

Anything that was shot with truely analog equipment (and most DV and SD-TV sources) sticks to the real analog scanline length of 52 µs which equals 702 active pixels. Since in the process of sampling to digital we're forced to add the missing half lines at the beginning and at the end of the active analog picture to make up full lines (575->576 lines) the width can be incremented by the same relative amount (703.2) and slightly rounded up to 704 pixels. 704x576 = 4:3, so PAR is 12:11.

[usually_quiet]

Anything that was shot with truely analog equipment (and most DV and SD-TV sources) sticks to the real analog scanline length of 52 µs which equals 702 active pixels. Since in the process of sampling to digital we're forced to add the missing half lines at the beginning and at the end of the active analog picture to make up full lines (575->576 lines) the width can be incremented by the same relative amount (703.2) and slightly rounded up to 704 pixels. 704x576 = 4:3, so PAR is 12:11.

Lovely, but the OP captured at 720x576 4:3 DAR, and the PAR for the capture is 16:15.

The capture includes 4 pixels of overscan on the left, 12 pixels of overscan on the right and about 8 pixels of noise at the bottom that the OP doesn't want to see. Because he also doesn't want to see black borders around the video for some reason, now he has to worry about PAR, DAR and SAR calculations so he can crop the video to 704x568 and have it come out proportionally correct instead of masking the junk and leaving the resolution alone.

I don't get it. If seeing letter boxing, pillar boxing, and once in a while window boxing or untidy edges on the picture bothered me that much I'd have to stop watching TV and movies.

[jagabo]

Lovely, but the OP captured at 720x576 4:3 DAR, and the PAR for the capture is 16:15.

The OP captured with a DV capture device which follows the rec.601 specification. The pixel aspect ratio is 12:11. The full 720x576 frame is wider than 4:3.

[Skiller]

Lovely, but the OP captured at 720x576 4:3 DAR, and the PAR for the capture is 16:15.

Lovely, you just demonstrated your ignorance. He did not. No one can. Unless you manage to change an ADC's internal timing parameters to non-standard values (i.e. not sticking to ITU-R BT 601). The sheer fact that the total amount of black pixels to the left and right is >=16 is proof. Those black pixels aren't overscan they are part of the horizontal blanking interval, because in the analog realm the picture is supposed to be exactly this wide (702) and no more. The additional black pixels are excess space to allow for some safety margin if the picture gets shifted horizontally, but they are not part of the area that defines the 4:3 picture and therefore shouldn't be treated as if they were.

He can just crop whatever he wants (without resizing) and set the PAR to 12:11 when muxing to MKV. What's so special about that?

[usually_quiet]

Thanks for finally providing an decent explanation. I cannot recall anybody in this forum ever mentioning that the pixel aspect ratio is fixed for DV. All one ever sees is SAR = Horizontal Resolution / Vertical Resolution. and PAR = SAR/DAR. For being a dick about it, bite me.

[freee]

OK, thanks to all gurus who contributed to my post. I have just realized that I have another option to get better result provided my logic is right. I was going through the menus of my Yamaha RX-V861 Receiver and bumped into the following option.



If I am reading this correctly I can:
* Connect my Panasonic NV-HD670 VCR to the receiver (composite to composite)
* Set the conversion on the receiver to composite to s-video
* Connect s-video out on the receiver to my old Panasonic DV Camera (it has s-video in input) and use the dv pass-through functionality to re-capture my tapes in dv-avi. Tapes are really in good condition. Or, use my old analog TV card PV951 to capture them as raw avi's (the card has s-video in).
As I stated in my original post I have captured my VHS and VHS-C tapes in dv-avi years ago but I am not very happy with the quality. I have come across many suggestions that s-video is the way to go but the required hardware is either not available or prohibitively expensive.
I am from a PAL land. I also have Canon HV30 but it has no s-video in option.
Am I really going to get a better results by using s-video?

[jagabo]

If I am reading this correctly I can:
* Connect my Panasonic NV-HD670 VCR to the receiver (composite to composite)
* Set the conversion on the receiver to composite to s-video
* Connect s-video out on the receiver to my old Panasonic DV Camera (it has s-video in input) and use the dv pass-through functionality to re-capture my tapes in dv-avi.

Whether that works better will depend on whether the receiver has a better converter than the DV camcorder.

[Skiller]

You can use the Yamaha as a passthrough but it's not guaranteed this is will improve things, actually chances are it's making it worse.

It really comes down to this:

• is the comb filter which separates the Composite signal to S-Video better than the one in your camcorder?
• how well does the Yamaha stabilize line jitter and how does it handle more severe signal dropouts (between recordings for example)?
• does it clip or alter the video levels?
• does it add noise or does it do some undesireable processing?

You can only find out by testing.


With VHS the advantages of a raw S-Video output (from an S-VHS deck) are only small. Once it's composite it depends on the comb filter in the connected device. The quality can be anything between rather much rainbowing and dot-crawl to almost S-Video like.

[newpball]

Am I really going to get a better results by using s-video?

If your original signal quality is not s-video but composite you cannot "make" it s-video quality afterwards.

Also why you would want to recapture through a camcorder is beyond me. Just buy, borrow, rent, whatever a good capture card and save the results uncooked in a lossless format.

[Skiller]

If your original signal quality is not s-video but composite you cannot "make" it s-video quality afterwards.

Yes, but since at some point every Composite signal needs to be separated into Luma and Chroma (either for capture or for display), having the separation well done is making a difference. Say the Yamaha does a really good job of Y/C separation and his camcorder does not. In that case there would be an improvement from using S-Video.
Of course with VHS the optimal way would be to use an S-VHS deck and use it's S-Video output since Luma and Chroma are already stored separately on the tape (yes, VHS stores them separately).

[newpball]

Say the Yamaha does a really good job of Y/C separation and his camcorder does not. In that case there would be an improvement from using S-Video.

And if we get 1001 other problems we solve these by perhaps putting another device in-between!

What are we talking about, we are going from:

VHS-C -> VHS -> Camcorder -> DV

Okay the VHS-C tapes are lost so why would anyone not recommend to go from VHS -> good capture card -> lossless codec?

[Cornucopia]

@newpball, there are a number of ways to convert from VHS->digital. How one goes about it depends on their budget, workflow and expectations.

If the OP is already USED to working with DV material, and if the OP already HAS DV conversion equipment, I think that's a very compelling reason to have him keep doing it that way (assuming Y/C and timebase problems are covered). It has been demonstrated that, for many consumers who don't intend to do much (or any?) processing/effects, DV is more than sufficient to capture VHS's quality.

That's clearly not true with every VHS out there, nor every user, and if one is more discriminating and intends to do a reasonable amount of corrective/enhancement/NR/processing work, it makes much better sense to go the route you are suggesting (in fact I would suggest that also).
But it's a matter of appropriate tool for the job and for the worker/audience.

Scott

[freee]

Say the Yamaha does a really good job of Y/C separation and his camcorder does not. In that case there would be an improvement from using S-Video.

And if we get 1001 other problems we solve these by perhaps putting another device in-between!

What are we talking about, we are going from:

VHS-C -> VHS -> Camcorder -> DV

Okay the VHS-C tapes are lost so why would anyone not recommend to go from VHS -> good capture card -> lossless codec?

I thought I mentioned that I have located the majority of the original VHS-C tapes and that they are in really good condition. My VHS-C camera is dead but I can play the tapes in my VCR with the VHS-C adapter.

Given my limited experience with VHS capture I resorted to asking here for your comments in relation to any potential benefits in using composite to s-video conversion via Yamaha receiver. It appears from your comments that there will be none, perhaps some or it may possibly have a detrimental effect on the capture if my Yamaha receiver does not do a good job.

The bottom line is I can only work with what I have and I will not be buying a new capture card because I have two already (Phillips saa7134 and one bt878).

So, I will conduct two testing scenarios and perhaps post the results here:
First - no s-video conversion
* Connect Panasonic NV-HD670 VCR with original VHS-C tape in it to Panasonic DV Camera (composite to composite)
* Use the dv pass-through functionality to capture the tape in dv on my PC (Firewire)
Second - with s-video conversion
* Connect Panasonic NV-HD670 VCR to the receiver (composite to composite) with original VHS-C tape in it
* Set the conversion on the receiver to composite to s-video
* Connect s-video out on the receiver to the Panasonic DV Camera's s-video input (it has s-video in input) and use the dv pass-through functionality to capture the tape in dv.

These videos are very important to me I am quite happy to spend time working on them.

[newpball]

These videos are very important to me I am quite happy to spend time working on them.

In that case I strongly advice you to buy a good capture card, then capture uncooked and save the results lossless.

[jagabo]

Regarding the receiver pass through vs. direct connection issue: You might as well try both and see which works better.

[freee]

I have completed the testing and my untrained eye cannot find any obvious difference between composite and s-video file. Captured as per above mentioned scenarios and converted to mkv's. No pre or post processing applied. The source is one of my VHS-C tapes that I regard to be in worst condition. Other tapes are almost perfect and I expect much better outcome.

[LMotlow]

Hmm, hard to say. We don't have a sample of your capture. What we have are re-encoded samples. Why are they encoded as progressive? Why is the frame size 720x572? What we can see in the re-encoded versions is that they appear to be capped with the wrong IRE (black levels too high and brights blown away with invalid luma levels) and there are line sync errors. Did you want comments about your captures, or about the way they were encoded with Handbrake?

[jagabo]

Yes, the original caps would be better for comparison.

The most obvious difference is in the levels. The composite capture has very blown out brights. So does the s-video capture, but not as bad. I adjusted the levels to match, and brought them into the legal range, for these screen shots:





Look at the top right corner. With the composite video you can't tell where the mountain ends and the sky starts. It's clearly visible in the s-video capture. You may be able to get around this by adjusting the proc amp settings in the capture device.

The composite cap has slightly more detail and slightly more color saturation. The s-video cap looks like it's been through a light noise reduction. The saturation can probably be adjusted with the proc amp. The loss of detail probably not -- cranking up the sharpeness might help.

Hint: copy the two images to your Desktop or some folder on your computer. Use Windows Photo Viewer to view them, using the left and right arrow keys to swap back and forth. Even tiny differences will become obvious. Use Windows' Magnifier to zoom in to see the smallest details.

Note the above images are JPG so they've lost a little detail. I meant them only for comparison of the blown out brights.

[LMotlow]

Looks to me like clipping was in the capture. But could be in the source. We need unprocessed caps to see what's going on. If you don't know how to make a cut from the original, let us know.

[jagabo]

Before I adjusted the gain on the two samples both waveforms went all the way to the top of the graph. I pulled the s-video sample down so that the max was at ~235 and then pulled the composite down so that the rest of the waveform matched. The OP should definitely play with the brightness/contrast settings on the capture device to see if he can bring the peak brightness down so it doesn't get clamped at 255.

[freee]

Looks to me like clipping was in the capture. But could be in the source. We need unprocessed caps to see what's going on. If you don't know how to make a cut from the original, let us know.

I will upload the Avi's tomorrow (I am away from home). They are around 150mb each. Yes, Handbrake was used for conversion. Just to reiterate, the captures were done in dv (analog pass-through via Panasonic DV camera). Conversion from composite to s-video via Yamaha receiver added to the chain . Composite capture, VCR - DV camera - PC (winDV capture utility). I did not change any setting s nor I could. I have two TV cards that I could use perhaps. If the Avi's files are too big, please let me know what's required.

[LMotlow]

The upload size limit is 500MB. The server is a bit slow (but it's free). Just give it time.

[jagabo]

You don't have to upload such long clips. Shorter clips with a mix of bright, dark, highly saturated colors, motion and stills would be better.