Questions about VHS Capture Project

[jagabo]

does those adjustments for the ati cards vary from model to model ?
from chipset to chipset ? from card to card ??

Yes. Card-to-card (same chipset and model) varies the least. The others may not even use the same range of values because of driver and chipset variations.

But you can't calibrate your caps to a DVD player and then assume the same settings are good for a VHS deck. You can calibrate caps from the VHS deck using a calibration tape - but that wont help either because the tapes you will be capturing probably weren't recorded on a calibrated recorder. So you pretty much have to eyeball every time...

[smartel]

and a calibration dvd taped onto a vhs ?

[jagabo]

and a calibration dvd taped onto a vhs ?

How do you know the DVD player output was to spec? How do you know the VHS deck recorded onto tape to spec? How do you know the output of the VHS deck is to spec? How do you know any other tape you play with the VHS deck was recorded to spec?

[jrodefeld]

I just didn't want anybody to get the wrong impression.

There's a lot of filters in the high-end S-VHS VCRs, and even the ones that "blur" only do so within a small range, and even then only to part of the signal (chroma or luma, rarely both). Some also appear to do temporal blending (Panasonic, not JVC). Some attack more complex signal aspect, such as chroma removal via TBC NR combo circuitry.

Lower-end Panasonic (like multisystem) had NR that just blurs. It's more like a sharpness knob on a TV, that can be cranked up or down. You can see it performing as such simply by watching the screen. I'd have to say these are more rare than not, mostly found on exotic VCRs that were not pro quality (or claimed to be and were not).

Consumer VCRs did the other way -- the "detail enhancer" junk that made tapes grainy. Like sharpness knob on steroids, and + values only. Butcher work.

I've not seen an AG-1970 to date, but you may be correct that this particular VCR blurs for an NR switch, like the AG-W3.

So, do you think I should set the Noise Reduction switch to on or leave it off when capturing? Or if you don't know about this particular model, on other higher end S-VHS players do you typically turn on NR hardware filters and so forth?

I haven't tested capturing with it off vs on, so I can't be sure what the differences are.

[jrodefeld]

I find it a little hard to believe that VHS caps don't need at least some filtering.

It's possible to pre-process enough via hardware to not need any software work.
I do it all the time.

At the same time, people do have a tendency to over-process digital video in software.

In your opinion, do you believe that it is preferable to process with hardware and do less software filtering? How do you process with hardware? And what role to you think there is for filtering with virtualdub, using things like NeatVideo and color correction and so forth?

Do you use virtualdub or avisynth for software filtering? When you do software filtering, what are your favorite filters or methods? If you could give a typical filter chain or method for a normal VHS capture, it would be helpful.

Also, speaking of hardware, do you consider the tools I have sufficient to do a good job? Where do you rank the Panasonic players in comparison to JVC and others? I am using a Time Base Corrector and the ATI TV Wonder card, but I am not using a Proc amp or dvd recorder or anything like that. Although, as others have said, the TBC has a rudimentary Proc amp where I can change brightness, contrast, color, etc. Do you advise that I use those controls?

I am not really interested in buying any new equipment, I just am curious as to how the hardwhere I have compares to what you would consider to be ideal for VHS capturing.

Thanks for all your help.

[jrodefeld]

I have another question, specifically for LordSmurf, though anyone who can answer would be very appreciated. I looked at the link that LS provided that said that CCE is NOT the best encoder and Procoder and MainConcept are much better.

http://www.digitalfaq.com/forum/video-conversion/2534-cinema-craft-encoder.html

I have been looking at MainConcept Reference, which seems to be very popular and an industry standard of sorts. The problem is I cannot find a way to do more than two pass VBR encoding. Is this a limitation of this software? I have heard that there is benefit to doing three or four pass encoding yet it would seem that a limitation keeping it at two pass would be a significant knock against MainConcept.

For anyone who likes this software, is there a way to enable multipass encoding? And if not, why do you not think this is a problem with this software?

And to LordSmurf specifically, in an earlier post you suggested that either Procoder or MainConcept were better than CCE. But which Mpeg2 encoder is your absolute favorite?

Thanks a lot for all the great answers I have received so far. This is really an invaluable resource for learning.

[jrodefeld]

As a follow-up to my previous post, I looked through the forum and learned that LordSmurf apparently is a big fan of Procoder. So I have some questions:


1. Which version of Procoder is the best? The latest version is 3.0. Most of the previous threads I read stated that 1.x was the best, while 2.0 had some problems. None mentioned 3.0, probably because it hadn't come out yet when the comments were written.


2. You mentioned elsewhere that you liked TMPGEnc when you wanted to use filters to restore the video. What filters do you use? Also, Procoder has some filters that can be used as well, as does MainConcept. Why not use those?


3. Can you frameserve from virtualdub to Procoder?



Thanks.

[jrodefeld]

I have one more question for anyone that cares to answer. On several occasions on this thread and elsewhere, people have talked about doing multiple "runs", meaning applying certain virtualdub filters on their own and then others later. I am a little confused about this. My process has been to capture Huffyuv AVI, edit with "direct stream copy" and then apply a filter chain and frameserve to an mpeg2 encoder only once. Maybe I don't understand but doesn't applying one filter, saving the avi and then applying others one at a time degrade the quality, sort of like reencoding video? That's why I have always thought that you should apply filters and encode in one step.

Could someone tell me why it is important to apply some filters one at a time? Thanks.

[sanlyn]

I have one more question for anyone that cares to answer. On several occasions on this thread and elsewhere, people have talked about doing multiple "runs", meaning applying certain virtualdub filters on their own and then others later. I am a little confused about this. My process has been to capture Huffyuv AVI, edit with "direct stream copy" and then apply a filter chain and frameserve to an mpeg2 encoder only once. Maybe I don't understand but doesn't applying one filter, saving the avi and then applying others one at a time degrade the quality, sort of like reencoding video? That's why I have always thought that you should apply filters and encode in one step.

Could someone tell me why it is important to apply some filters one at a time? Thanks.

To answer your last question first: many people do apply all their filters at the same time, then frameserve to an encoder. But many don't, for several reasons.

What if you applied a filter, but later you don't like the results? You have to repeat the entire process - assuming you saved your filter settings or kept notes on what you did. Of course there are times when you might have to re-process from scratch for some reason, even if you broke the project into separate steps.

Many people denoise and then correct color in separate steps, but they don't encode each step. They just save each step as a separate AVI. Let's say you start with an unprocessed AVI that you call version A. You'd apply some denoising and then save that version A as AVI version B. Then you load version B into AviSynth/VirtualDUb or whatever, apply other filters, and save that as version C (or frameserve to an encoder if you're satisfied). Because denoising almost always takes longer than color filters, people separate those operations. If you don't like the color or black levels, at least you don't have to denoise all over again.

No, working with AVI and then saving and re-using it for the next step doesn't degrade if you use lossless compression or no compression. But if you're working with a lossy compressor like DivX, you're in trouble. If you have to run a lot of RGB corrections, save those steps in RGB until you have what you want; then, if you want to convert RGB to YUY2 for your CCE encoder, don't do it until you're satisfied with your work in RGB or whatever colorspace you had to use.

Many like myself maintain that running denoisers and color filters at the same time hampers the denoiser. That might not be true in all cases depending on the filter, but I think it's a waste of time to denoise all over again if you don't like what you did with the color.

[poisondeathray]

I have another question, specifically for LordSmurf, though anyone who can answer would be very appreciated. I looked at the link that LS provided that said that CCE is NOT the best encoder and Procoder and MainConcept are much better.

http://www.digitalfaq.com/forum/video-conversion/2534-cinema-craft-encoder.html

You have to PM him specifically if you want him to reply

I have been looking at MainConcept Reference, which seems to be very popular and an industry standard of sorts. The problem is I cannot find a way to do more than two pass VBR encoding. Is this a limitation of this software? I have heard that there is benefit to doing three or four pass encoding yet it would seem that a limitation keeping it at two pass would be a significant knock against MainConcept.

Yes you can enable 2pass in the advanced options, or there is a checkbox in the simple MPEG2 configuration. What version are you using?

In general, 2passes or greater depends on how the software functions. Most software do not benefit from >2 pass (most software do not allow for >2 pass). CCE is the only one that comes to mind that does allow for >2 pass and actually has a demonstratable benefit when using >2 passes on some types of content

A lot of what you are hearing is opinion and experience (which is perfectly valid). The fact is, all MPEG2 encoders have strengths and weaknesses, and there are many areas or criteria you can examine. You cannot say one is absolutely better than another, and note there is a large degree of subjectivity in this assessment as well. There are objective measures, like PSNR and SSIM, but they do not necessarily have a high coefficient of correlation with subjective perception of "quality"

The MPEG2 encoder featured MCR actually isn't an "industry standard" and not used in any big studios (not necessarily because of lack of quality, it's actually quite good most people will agree) - it doesn't offer high end features like segment level encoding - that's a big limitation for studio level encoder.

[lordsmurf]

Certain errors have to be processed in hardware, else they become permanent. Software cannot fix.

I maintained that MainConcept is an industry standard software MPEG-2 encoder. when you look at everything -- not just the "blockbuster" movies in movie theatres.

Procoder 1 is not really worth discussing. Old, unavailable to buy, inferior to MainConcept, etc.

There's really no reason for more than 2-pass encoding. Those that do allow it mostly just make up for their own weaknesses in analysis.

VirtualDub color corrections work, but are wholly inferior to Adobe Premiere Pro. Inversely, Premiere NR is a joke, and VirtualDub and/or Avisynth is what you'll need there.

MainConcept and Procoder filters are lousy. The end.

TMPGEnc Plus filters are pretty decent, still useful, cheap -- but you can do better with a combination of Premiere Pro CS4, VirtualDub and Avisynth now. Of course, that has one hell of a learning curve, and requires more passes and more apps. TMPGEnc Plus is more newbie friendly.

Again, no PM, dumb luck that I saw this.

[jagabo]

TMPGEnc Plus filters are pretty decent, still useful, cheap

One problem with TMPGEnc Plus is that it does all its filtering in RGB after a rec.601 conversion. So you have no ability to recover super darks and lights from a YUV source.

[jrodefeld]

Certain errors have to be processed in hardware, else they become permanent. Software cannot fix.

I maintained that MainConcept is an industry standard software MPEG-2 encoder. when you look at everything -- not just the "blockbuster" movies in movie theatres.

Procoder 1 is not really worth discussing. Old, unavailable to buy, inferior to MainConcept, etc.

There's really no reason for more than 2-pass encoding. Those that do allow it mostly just make up for their own weaknesses in analysis.

VirtualDub color corrections work, but are wholly inferior to Adobe Premiere Pro. Inversely, Premiere NR is a joke, and VirtualDub and/or Avisynth is what you'll need there.

MainConcept and Procoder filters are lousy. The end.

TMPGEnc Plus filters are pretty decent, still useful, cheap -- but you can do better with a combination of Premiere Pro CS4, VirtualDub and Avisynth now. Of course, that has one hell of a learning curve, and requires more passes and more apps. TMPGEnc Plus is more newbie friendly.

Again, no PM, dumb luck that I saw this.

Thanks for the reply. I have looked through the forums here and there are several times that you have stated your preference for Procoder over other encoders. Has this changed? Not specifically Procoder 1.0 as I would expect that one to be outdated. But what about Procoder 3.0? Is that one better than Main Concept Reference? I just want to clarify which encoder and which version you use.

[sanlyn]

Nobody has even mentioned resolution yet? 720x480.

For 4:3 standard def., 704x480 lets you keep a little more of the full original frame. 720x480 and 352x480 will cut pixels off the top and left. 704X480 frame aspect ratio is closer to 4:3 than the other formats.

[jagabo]

Nobody has even mentioned resolution yet? 720x480.

For 4:3 standard def., 704x480 lets you keep a little more of the full original frame. 720x480 and 352x480 will cut pixels off the top and left. 704X480 frame aspect ratio is closer to 4:3 than the other formats.

The DVD spec refers to the MPEG 2 spec as far as aspect ratios are concerned. The MPEG 2 spec clearly states that the full frame constitutes the indicated DAR. BT.601 specifies the 704x480 frame constitutes the 4:3 DAR and the extra 16 pixels, if present, are for padding. Exactly which spec do DVD players and DVD producers follow? It seems to be a mix of both. And it can vary depending on whether you use the analog of digital outputs.

[2Bdecided]

I find it a little hard to believe that VHS caps don't need at least some filtering.

It's possible to pre-process enough via hardware to not need any software work.
I do it all the time.

The noise reduction available in software is far more powerful (i.e. removes more noise while maintaining more details) than anything you'll find in consumer hardware.

Not all tapes need it of course.


Some of the fixes for noisy lo-res chroma, and halo removal, etc, are almost like magic. Though they can be abused.

Cheers,
David.

[2Bdecided]

For 4:3 standard def., 704x480 lets you keep a little more of the full original frame.

For any standards-compliant capture device, that's not true.

Original NTSC was equivalent to approx 711x486. No consumer equipment gets those extra 6 vertical lines into the digital domain (and few analogue broadcasts used them for the last few years), but 720 vs 704 does exactly what it says: gets the full 711 plus a little extra blanking.

704x480 is a smaller 4x3 window within the larger original 711x486 4x3 window. I'd use 704x480, but it's not correct to say it preserves more of the original frame than the wider 720x480.

Cheers,
David.

[Nelson37]

I didn't use any equipment to "calibrate" the card, because I didn't have any.

My reasoning was that the capture, played back thru the same device as the source, should be as true to the source as possible. Whether the DVD playback was "correct" or not really shouldn't matter, if the capture matches the source.

They were consistently too dark and the color was off a bit. Once this was adjusted, the captures were true to source, which I tested with different sources.

Different cards and different software varied this slightly, but not all that much.

[jagabo]

My reasoning was that the capture, played back thru the same device as the source, should be as true to the source as possible. Whether the DVD playback was "correct" or not really shouldn't matter, if the capture matches the source.

Yes, but you've only calibrated to that particular devices analog output. The settings aren't useful for anything else except as a rough starting point.

In my experience the analog output of different devices varies quite a bit. So I adjust the settings for each device. And with analog media (VHS, for example) what's recorded on the media can vary. So I adjust to the particular recording.

[sanlyn]

TMPGEnc Plus filters are pretty decent, still useful, cheap

One problem with TMPGEnc Plus is that it does all its filtering in RGB after a rec.601 conversion. So you have no ability to recover super darks and lights from a YUV source.

But...serious question: are you saying that TMPGenc 2.5 should do its filtering in, well, which other color storage technology (sRGB?), and then convert to Rec.601 later? What would the difference be? Do Cinemacraft and other encoders encode to sRGB or something other than Rec601/709, etc.? I thought for years that I had these video color gamuts and standards figured out, but I think I'm getting confused.

If I record my own DV digital video as YUY2 or YV12 with some nice super darks and super whites, convert it to AVI for a little color correcting (and let's say I don't use an RGB app like VirtualDub or Adobe Premiere or AE, let's say I just use ColorYUV in AviSynth to keep Aunt Jessie from looking purple)...then I send that AVI in its original YUV to Procorder or CCE or whatever, and I make an NTSC or PAL DVD out of that AVI. What would the color differences be between the YUV where I started and the YUV where I ended? What would happen to the original super darks and super whites?

[jrodefeld]

Hello everyone,


I have another question that is changing the subject slightly. I was browsing eBay and stumbled upon a Vidicraft Vidimate VDM 300S and promptly bought it. Now, Vidicraft detailers are fairly well respected around here from what I have gathered and can seemingly do wonders for cleaning up VHS tapes. From what I have read the Vidimate series is a sort of combination detailer and Proc Amp and beyond that the 300S is the only Vidicraft unit to have S-Video inputs and output.

So, a detailer and Proc Amp from a respected brand WITH S-Video inputs for $45? I couldn't pass this deal up. Even if it doesn't help much I am not out much money so it is certainly worth a try.

So, could anyone tell me about the application of this type of unit in capturing VHS? What function does a detailer provide and what do you know about the Proc Amp functions on the Vidimate series? Now the TBC I have (Cypress CTB-100) has a weak Proc Amp but I assume I should bypass this in favor of the presumably more sophisticated Proc Amp features on the Vidimate 300S? With this added to the equation, my hardware chain looks like this:

Panasonic AG-1970P, S-Video cable to Cypress CTB-100 TBC, S-Video Cable to Vidicraft Vidimate 300S, S-Video Cable to ATI TV Wonder 650 Capture Card.

Now that should provide a decent amount of hardware filtering and I believe I should be able to do more than an adequate job in capturing my VHS home movie collection.


My questions are:

First, what do you know about this detailer (and Proc Amp), the Vidimate VDM 300S? Is this a good purchase for less than $50? How much improvement can I expect to see with this unit incorporated into my capturing process?

Second, will the detailer and color correction controls allow me to bypass a good deal of software color correction and filtering? I am just curious to know what I can expect to achieve in hardware vs software with such a setup.


Thanks everyone.

[poisondeathray]

TMPGEnc Plus filters are pretty decent, still useful, cheap

One problem with TMPGEnc Plus is that it does all its filtering in RGB after a rec.601 conversion. So you have no ability to recover super darks and lights from a YUV source.

But...serious question: are you saying that TMPGenc 2.5 should do its filtering in, well, which other color storage technology (sRGB?), and then convert to Rec.601 later? What would the difference be? Do Cinemacraft and other encoders encode to sRGB or something other than Rec601/709, etc.? I thought for years that I had these video color gamuts and standards figured out, but I think I'm getting confused.

If I record my own DV digital video as YUY2 or YV12 with some nice super darks and super whites, convert it to AVI for a little color correcting (and let's say I don't use an RGB app like VirtualDub or Adobe Premiere or AE, let's say I just use ColorYUV in AviSynth to keep Aunt Jessie from looking purple)...then I send that AVI in its original YUV to Procorder or CCE or whatever, and I make an NTSC or PAL DVD out of that AVI. What would the color differences be between the YUV where I started and the YUV where I ended? What would happen to the original super darks and super whites?

That's exactly the issue with Rec601/709 . This means Y' 16-235 => RGB [0,0,0] - [255,255,255] . This means Y' values <16 (superdarks) , Y' >235 (superwhites) are clipped. Bye Bye.


So the options to salvage these are :

1) Legalize the Y' (and CbCr) values before converting to RGB . ie. bring the supervalues back into Y' 16-235 range, and CbCr 16-240 range

2) Convert to RGB using full range. Full range means Y' 0-255 => RGB 0,0,0 - 255,255,255

3) Use software that can access Y'CbCr before RGB conversion and filters - some software can access Y'CbCr colorspace before the RGB conversion and filters this allows you to salvage the supervalues within the program. e.g. Premiere can do this with some filters and certain filetypes that it "treats" as Y'CbCr. eg. You would be able to do this with DV, but not Huffyuv (because PP doesn't treat Huffyuv as YUV) . I don't want to get into specifcs - Lets just say each software has quirks and you have to learn them if you want to get the most of them.

[lordsmurf]

some software can access Y'CbCr colorspace before the RGB conversion and filters this allows you to salvage the supervalues within the program

Or ....... Avisynth into TMPGEnc.

[sanlyn]

Oh. Then what I understand about color/level ranges and what poisondeathray and lordsmurf are saying are the same thing. I do keep an eye on 'legal' RGB values when processing - but trying to correct many color problems in YUV alone is a real hassle, if not impossible (you get the purple out of Aunt Jessie, but it makes Uncle Frank look green). When I carelessly encode out-of-range color into DVD, TVs that display them behave oddly anyway. I use TMPGenc 2.5 for lots of color work (I think the old 2.5 and "Plus" are two different encoders, I didn't care for Plus), but I watch its histograms carefully. This doesn't explain why I've recorded downscaled HD shows into a Rec601 DVD Recorder (Toshiba RD-XS34), copy them to my PC and convert to AVI with a Rec601 matrix, and still see 0-255 all over the place.

But that's a story for elsewhere . . .Sorry for the detour.

[sanlyn]

So, a detailer and Proc Amp from a respected brand WITH S-Video inputs for $45? I couldn't pass this deal up. Even if it doesn't help much I am not out much money so it is certainly worth a try.

Vidicraft made good stuff. I hear that the same engineers moved on to have a hand in the design of the current line of proc amps at SignVideo.

So, could anyone tell me about the application of this type of unit in capturing VHS? What function does a detailer provide and what do you know about the Proc Amp functions on the Vidimate series? Now the TBC I have (Cypress CTB-100) has a weak Proc Amp but I assume I should bypass this in favor of the presumably more sophisticated Proc Amp features on the Vidimate 300S?

Good equipment, but learning curve required. Very similar to the "200S" model. Discussion here:
https://forum.videohelp.com/threads/232027-What-type-of-unit-is-the-Vidicraft-Vidimate-...=1#post1356084

. . .will the detailer and color correction controls allow me to bypass a good deal of software color correction and filtering?

A good deal of it. Not all. If you're satisfied with the results, it's in the bag.

I am just curious to know what I can expect to achieve in hardware vs software with such a setup.

Now that you have all the pieces in place, there's only one way to find out.

[sanlyn]

For 4:3 standard def., 704x480 lets you keep a little more of the full original frame.

For any standards-compliant capture device, that's not true.

. . .

704x480 is a smaller 4x3 window within the larger original 711x486 4x3 window. I'd use 704x480, but it's not correct to say it preserves more of the original frame than the wider 720x480.

I'll take your word on that, David. Maybe it's a case-by-case basis? I capture VHS to 640x480 AVI in VirtualDub. If I encode to 720x480, I lose about 15 pixels on the left (the image is shifted left) and a few at the top. At 704x480, I lose less.

[2Bdecided]

I'll take your word on that, David. Maybe it's a case-by-case basis? I capture VHS to 640x480 AVI in VirtualDub. If I encode to 720x480, I lose about 15 pixels on the left (the image is shifted left) and a few at the top. At 704x480, I lose less.

I thought you meant when capturing, hence my comments. If you're converting from 640 to 720 or 704, then what you get depends entirely on the conversion. In AVIsynth, I could make anything happen

Cheers,
David.

[jagabo]

This doesn't explain why I've recorded downscaled HD into a Rec601 DVD Recorder (Toshiba RD-XS34), copy them to my PC and convert to RGB32 AVI with a Rec601 matrix, cut/edit the AVI in VirtualDub, make fades, fix audio, or whatever, etc., etc., with no color adjustments but still in in RGB, noting all the while that histograms are showing me 0-255 in there -- then send that AVI to TMPGenc and into DVD, and neither see nor measure any difference in 0-255; the same range I started with is where I finish.

Because the YUV<-->RGB conversions are symetric. The luma range is expanded when converting from YUV to RGB, compressed when converting from RGB to YUV. The problem is the super darks and super brights are lost when converting from YUV to RGB. Ie, all Y values from 0 to 16 become RGB=0. Then when converting back to YUV they all become 16. So any details in those dark regions are lost.

Open the DV AVI file in this post:

https://forum.videohelp.com/threads/326496-file-in-Virtualdub-has-strange-colors-when-o...=1#post2022085

with TMPGEnc Plus. No adjustments of its filters will bring out the bars below y=16 or above y=235.

[sanlyn]

Thanks, jagabo, what you describe is what I've always understood. It's that expansion/conversion I'm always careful about, and go thru a lot of trouble/preview/rework to place those "details" in the right place before I encode.

[poisondeathray]

some software can access Y'CbCr colorspace before the RGB conversion and filters this allows you to salvage the supervalues within the program

Or ....... Avisynth into TMPGEnc.

There can be few "gotchas" when frameserving into other programs. Sometimes the frameserver does the RGB conversion (Ie. it doesn't present Y'CbCr to the 2nd program, and the 2nd program cannot access Y'CbCr in the script). This is a "double whammy" , because not only are "superbrights/darks" clipped, the interlaced chroma sampling for YV12 sources is done incorrectly because it's presumed to be progressive upsampling. I haven't tested specifically with TMPGEnc, but this can occur frameserving into other programs . I think it's generally good advice to conduct some test and learn the quirks of whatever workflow/software you plan to use