"Professional" Video Capture

[2ZOD.COM]

.....

[trevlac]

I don't have such a device, but you're source probalby makes a big difference.

'Professional' equipment would not have a comb filter because it would be component. Also, the connectors would be different than what you might expect. Finally, the pro devices seem specific to a specific job, so their other functionality, like a comb filter on a TBC might actually be bad.

Here is a link to some professional ADCs. http://www.bhphotovideo.com/bnh/controller/home?O=NavBar&A=search&Q=&ci=6240

Here is a link to the Matrox RTX.100 http://www.bhphotovideo.com/bnh/controller/home?O=productlist&A=details&Q=&sku=279383&is=REG

It's really ment for cap and editing. It's pro-sumer and more likely what you would want.

I don't really know this stuff. This is just what I've read.

[2ZOD.COM]

.....

[ppera2]

I think that it's simple impossible to get even near DVD quality from VHS tapes. Resolution is too low - that's the main problem.
Professional capturing may help by damaged tapes (TBC), noise filtering, but resolution will be still low, even by S-VHS too.

BT8x8 chips are cheap, but that doesn't means that they are insufficient for VHS. Actually, they are too good for VHS in some manner - bandwith is much bigger. Usually lowering sharpness gives much better results by VHS capturing.

[2ZOD.COM]

.....

[ppera2]

Originally posted by 2ZOD.COM
.........There was really no point to this post, just kind of ranting.

Sorry bud, but you really should care much more what and how write here...
After writing about BT8x8 and integrated cards (very general, and shallow) you talk about some ranting?
After nonsenses like: "capture 800x600", "...VCR's that can convert directly to a digital signal, so there is no loss..." .
Any A/D conversion is inperfect, and always will be some loss.
I could talk much more about, but why?

[2ZOD.COM]

.....

[Arachnotron]

@2ZOD.COM

Sorry to intrude, I couldn't resist :)

That wasn't really anything, I just read somewhere about the Philips SAA7108E being able to capture a theoretical resolution of 800x600

The 7108 is mainly used... to slap onto an AGP video card :)

It is, like most Philips chips in the 71xx series, running at 27 MHz. That makes it's theoretical maximum resolution about 52.66 X 27/2 = 710 pixels in the standard NTSC window (or 720 in the 53.333 µs DVD window)
The BT878 works at a higher clock, and could theoretical go higher!

But this is a theoretical maximum; due to various low pass filters the maximum usefull resolution will be lower than that. (about 640 for the BT878, somewhat higher for the Philips)

Some Philips drivers will let you cap at higher resolutions. A SAA7134 based card I own would even let me do 1000x800. Completely pointless, since this is basically a resize from 710 pixels.

So, look at the sampling rate when judging cards first. Cap resolution doesn't mean a thing!

Your VCR idea is not as crazy as it sounds. Many higher end S-VHS VCR's already operate competely digital, since things like comb filtering (separating the chroma from luma in composite video) can better be done digitally. The same goes for high end TV's. This leads to the crazy situation that when you play back an analogue VHS tape to an analogue TV, the signal has gone through two digitizing cycles.

Now wouldn't it be nice if you could simply plug an USB/firewire cable in such a VCR... :)

[2ZOD.COM]

.....

[IanD]

Originally posted by Arachnotron
Some Philips drivers will let you cap at higher resolutions. A SAA7134 based card I own would even let me do 1000x800. Completely pointless, since this is basically a resize from 710 pixels.

Arachnotron,

Do you know if VFW drivers are available anywhere for an SAA7134 based capture card?

I have a Flyvideo 3000, but am stuck with a VFW system and the VfwWdm wrapper is not stable enough on my system to reliably capture more than a few minutes.

Philips definitely have VFW drivers for the SAA7134 via an SDK, but refuse to release to the public: insisting that the manufacturers support the customer (which of course they are only interested in doing via WDM drivers).

Ian

[Arachnotron]

Do you know if VFW drivers are available anywhere for an SAA7134 based capture card?

I don't know, sorry. I never came across them, but I never really looked fro them in the first place. I only use directshow capture apps, so the problem never came up for me.

But I thought vfw cap in w2k and XP was always emulated? Please correct me if I am wrong, because I am not 100% sure about this.

[valnar]

I use a Prolink Pixelview Xcapture card which has the Conexant CX23881 chip. This is much better than their older BT878 chip. The colors are more vibrant, and definitely above the resolution of VHS. For keeping audio in sync, use a program like VirtualVCR from www.digtv.ws. Nothing else works as well.

It only supports s-video and composite though, but the card is very cheap.

If you wanted something that captures with locked audio/video, and converts to DV on the fly (therefore, to firewire on the PC), I've heard nothing but good things about the Canopus ADVC100. That's the ticket I think you want. Would 720x480 D1 (NTSC DVD) resolution be good enough?

Perusing the Canopus Corp web page, it looks the like the ADVC300 is even a higher end version of the ADVC100. It might be worth a look.

http://www.canopus.us/US/products/ADVC-100/pm_advc-100.asp
http://www.canopus.us/US/products/ADVC300/pm_advc300.asp

-Robert

[echooff]

I recently purchased the Canopus ADVC 100 and am quite pleased with it. I am capturing my old vhs home videos without any audio sync problems or dropped frames. The only problem is it doeesn't automatically edit the the stuff I don't want to keep:D :D :D Just kidding.

[FredThompson]

Originally posted by 2ZOD.COM
That's exactly what I mean. That would be one of the best things for video capture, the VCR would probably have composite inputs too so you can hook up a hi8 camera up to it too or something. Something like that would be so great, but would never be realized because it would cater to such a small group of people.

Yeah, I know 800x600 is completely useless, but I just always see all kinds of so-called professional stuff with just ridiculously high stats no matter what it is. Ya know? It's just another kind of pointless exaderation. There are dual-deck S-VHS/DV decks with firewire out that do this type of thing. Then you get to clean up the 4:1:1 color mess.

There is no such thing as perfect A/D nor is it possible to remove analog artifacts without also removing some of the information you want to keep. This isn't like washing mud off your shoes, noise overlays or changes the data which you want so it's a balancing act.

Your question and posts show you really should do more research.

[vhelp]

.
.
I also believe that for the "closer to source" look, you have to find
a card that will match as closer to the source's sweet-spot, or
resolution. No one really know for sure what the resolution is (if
there really is one, for that matter) but I also think we need to visit
the TV (say a standard NTSC, not ws) and do a little research as to
how many true lines of resolution a tipicle tv set is.

I understand your point, to match as close to the source as possible
w/ your VHS tapes. I too, have ben in the VHS conversion for a while
now, but am still learning some things here and there. Anyways..

As I was saying.. I often wonder why, when I play my tapes (commerical)
that they look so good (not dvd) but when I capture it (be it Analog
or DV) the source is noisy (assuming no noise from capture card or
mobo/chipset etc) only from the capturing intricicies. I often see
these pixels that do not match others nearby pixels, or look out of
place.

Another "illusion" is when you are capturing via Analog card, and you
have the ability to view it via Overlay window (ie, in vdub) When you
watch the VHS source in your Overlay window, it always looks very clean,
but soon as you capture it, it's dirty, and w/ plenty of out-of-place
pixels that don't seem to line up w/ nearby pixels.
.
.
Listen, if a TV set can pic up a VHS signal and dispaly it "clean"
weather it's by electronic circuitry or scaline technique, than an
Analog capture card should be able to do the same thing. I don't get
it, unless.. THEY don't want ot bother w/ such a limited source to
begin with. Actually, I don't think it's limited. I think THEY limit
it !!

Just my 2 cents worth.
-vhelp

[ppera2]

@vhelp

I don't agree that overlay preview looks better than capture. If capture looks bad, it's because of compression, which work very bad with noisy source. In such case you need to choose better quality or another codec (who said huffyuv? :) ).

2: Noise is much more visible on monitors, due to higher resolution.
TV pic. tube has much bigger raster size, and itself alone makes some 'noise filtering'.

3: A/D conversion works very bad with noisy source. It costs much electronic to properly filter out noise before A/D converting.
That's reason for higher price of pro. cards.

[Arachnotron]

The most important factor in quality whem capping tapes is not so much the capture device (though important). It's the VCR. No amount of fancy filtering will compensate for what is lost in a crappy signal produced by a cheap VCR.

The return in quality per dollar/euro invested in a good VCR is much larger then the return you get from a really expensive capture device.

Make sure you at least use a VCR that can produce a s-video signal, even when playing back ordinary VHS tapes.

[edit] and cables of course. Good quality, short shielded cables. And keep your PC case closed. And grounded. And.. and.. In short, make sure your device get's the best possible quality analogue signal first before complaining about a capture device.

[FredThompson]

Absolutely! GIGO. That's why I bought one of the higher-end S-VHS decks from JVC. The stabilization of the image and color due to it's filters is well worth the investment.

Another key is NOT trying to remove the appearance of "tape." Too often people filter away all the details and are excited because then they have very little swim in the color. Video processing is kind of like archaeology. If you want good results, you need to use a light brush, maybe a few different light brushes. If you just cut away with a bulldozer, you'll destroy much of what you want to save.

[vhelp]

@ Fred,

yes. I agree. In my VHS projects, I prefer not to use any
filtering. I feel it's better to maintain all pertinant data,
inclding noise (other than card/mobo/chipset) to retain whatever detail
is present in the video, after a capture process.

Also, it's about that time, that we seek another alternative to storage
medium.. That being, DVD disks :)
.
.
By going in this direction, you virtually do away w/ such filtering
in your encoding. No need to. Just utilize the best capture device,
and top it off w/ the level of skills you possess.. IOW, raise the
bitrate accordinaly, to your VHS-to-eye coordination to detail.
Some tapes require higher bitrate than others. IF vhs source is of
a widesceen nature, then you would take advantage of the AR and cropping
and process accordingly.

As far as professional devices.., well, I remember when the advc-100
first came out. At $300 + it was high-end (and most considered it of
a professional class) This is by far, one of the best devices so far,
that any user could aquire in their video endeavors. And, vhs is no
axception. The only issue(s) you have to attend to w/ DV is the color
space (aka, 411 chroma/luma sampling bug) and the external noise associated w/ obtaining video data through such devices (be it Analog
capture card or DV device) ..course, there are other issues w/ noise
and artifacts and such, but the above is the two most major of them
all.

Then, of course, there is the user skills level too. That has to be
factored as well. You can have the best setup, but if user is inexperience,
there no replacement for quality once completed (and distroyed)

Yes, a good VCR equipment is a must as well. I have a JVC S-VHS w/
s-video connecitn IN/OUT and I use them extensively in all my VHS projects.
I have an external TBC unit, plus a host of other exteran noise/color
devices to add as well - - all aquired through trial-n-error and picking
up knowledge over the years I embarked in video.

-vhelp

[Arachnotron]

I sometimes worry a bit about chaining devices and using filters in the VCR. By now, most of these are digital too. So your signal has gone through one or more analogue to digital to back to analogue conversions before finally being captured.

As I said before in this thread, it would be nice if you could siphon off the first digitizing step in the VCR and stay in the digital domain after that.

[FredThompson]

Depending on the content, I'll use either DV passthrough or an ADS Instant DVD 2.0 USB converter. DV does a horrible job with thin vertical lines. After some extensive testing and discussions, a couple of filters surfaced to help change 4:1:1 to 4:2:2. Reinterpolate411 is for AviSynth, 411Helper (I think that's the name, Xesdeeni wrote it) is for VirtualDub. Like anything else, they can't recreate missing data. However, they do help.

Anything with video bugs or sharp edges is horrible with DV. 4:1:1 just adds too much false motion. For stationary bugs, I'll typically copy it from a single frame after cropping everything else and using Donald Graft's averaging filter then use the logo filter to overlay which can help a lot. I'll also overlay any black bars if the source wasn't 4:3 or if it has scrolling junk at the bottom like the news channels.

If the source is laserdisc, it should be done via composite cables. If it is a PAL VHS tape, I also use composite because multi-format S-VHS is so expensive and I can't justify spending that kind of money on a near-dead video format. I do use Guava and have been monitoring the recent attempts to improve comb filtering.

I also use Peach quite a bit. That's about it. I AM testing various methods of removing scratches and dropouts. You can always check a thread in the DV forum to see what I'm using for non-test purposes.

Personally, I leave everything interlaced unless it's film or I'm trying to undo some of the cross-platform craziness. I can see no good reason to deinterlace and lose half the motion information and have artifacts from interlaced source.

Oh, VHS I always save as CVD. S-VHS is always saved as SVCD. Both have audio at 48. I realize CVD can look a little chunky on a large display. Proper post-filtering will take care of that.

I'm really hoping the current development of a GUI for AviSynth will incorporate a proper timeline. It would be wonderful to easily set when certain things should happen. For example, some captures might have letterboxed portions but the entire thing might not be letterboxed. It's far easier to determine when to start and stop visually. Sure, you can do it with VirtulaDub and a pad of paper but that's still messy.

Oh, I also always capture at the highest resolution the hardware allows. Not what the driver allows, what the hardware will do. I'd prefer to do the rescaling on my terms. After seeing some test that Ivo did, I've switched the processing order. Now it's format correction (comb filter, 4:1:1->4:2:2, remove chroma for B&W, etc.) then resizing then Peach. This appears to keep more detail than resizing as the last step.

[2ZOD.COM]

@FredThompson
<< There are dual-deck S-VHS/DV decks with firewire out that do this type of thing. Then you get to clean up the 4:1:1 color mess. >>
I didn't know if such things existed. It's so hard to find any information on specific topics such as that sometimes I just didn't know where to begin. I always like to think anything I think of has already been done, so I just thought I would ask rather than spend hours trying to find something.
I wouldn't doubt that that deck you speak of costs at least $2000. Which I don't have. I'm 16 and unemployed.

<< Your question and posts show you really should do more research. >>
Defineatly. I'm still VERY new to all of this. I've tried to figure things out by trial and error, but I don't even know what I should be trying sometimes. I'm very thankful for the doom9 forums to help me.

@vhelp & FredThompson
<< yes. I agree. In my VHS projects, I prefer not to use any filtering. >>
I agree as well. I do no filtering. Capture, convert to MPEG-2 and author to dvd. When someone wants me to convert their tapes to DVD I feel as though I'm just cutting out the detail when I try to filter out noise. Exactly what you said, and that's defineatly what I don't want.

[FredThompson]

If you're doing NTSC DV -> DVD, you really should use one of the colorspace correctors. Try the script I have over in the DV forum. It's the result of months of studying the process.

It's not that I think filtering is bad. On the contrary, I think it is necessary but too much is bad. Slight filtering can help reduce inconsistencies and increase the efficiency of compression.

The best use of money is one of the good higher-end S-VHS decks with stabilization and digital cleaning. They have access to the raw data from the tape. It's hard to explain how dramatic the difference is until you see it for yourself. You'll be shocked how much motion is in regular VHS as the image wanders around the screen.

[2ZOD.COM]

<< If you're doing NTSC DV -> DVD, you really should use one of the colorspace correctors. >>
So far all I've done is NTCS VHS -> DVD, so it doesn't really apply to me, but it is greatly appreciated. It's always good when there are people who invest a lot of time into something such as this where quality is of utmost importance.

I am defineatly going to get one of those decks with the firewire connection and internal digital filtering. With something like that it seems as though everything is easy as cake since it would be in a digital format. Capturing is such a tedious process with these bt8x8 cards if you're trying to achieve the best quality possible.

So then I assume you can capture using something like Adobe Premier and do all your editing with that when you have a firewire solution such as that?

Thanks for all your input, you have no idea how much I appreciate it.

[FredThompson]

I don't have a firewire S-VHS/DV box. They exist but are very expensive. I have a very good JVC S-VHS deck and capture via S-VHS to either DV passthre or an extrenal MPEG2 device. I prefer the MPEG2 box. The Brooktree device is one of the best available of that scale. You're probably not using it properly. Contact I4004 on another board and ask for configuration help.

[2ZOD.COM]

I found a relatively inexpensive deck with firewire in/out as well as a bunch of other options.
http://www.ggvideo.com/jvc/srvs30u.jpg
JVC SR-VS30U for $756.99 here (http://www.abesofmaine.com/viewproduct.asp?id=jvsrvs30u&l=CNET)
"IEEE 1394 In/Out of S-VHS/VHS or DV Decks"

I would say that's quite a reasonable price for it's functionality.

[FredThompson]

OK, so then you're pulling from the analog tapes as DV which is 4:1:1. Since you want to go to MPEG2 which is 4:2:0, you have to do something to correct the video. You'd be far better off to capture as 4:2:2 or 4:2:0.

You'd be far better off to get a standalone DVD recorder which can go at high bitrate and use standard DVDRW.

[2ZOD.COM]

Well my plan would be to capture using the firewire out from standard VHS, and whatever other sources I use in the future, capture in either uncompressed DV or HuffyUV, then convert to MPEG-2 most likely with tmpgenc (not definately) import to scenarist, burn to DVD.

The MPEG-2 compression is much better when it's not on-the-fly, or real-time. It also gives the flexibility of doing 2-pass VBR to fit more onto a DVD.

Any suggestions for what I'm doing?

[FredThompson]

Are you working with NTSC or PAL?

I didn't say to take the result of a high bitrate DVD recorder as your final product. I said you'd be better off to use a standalone DVD recorder in your situation.

[2ZOD.COM]

<< I didn't say to take the result of a high bitrate DVD recorder as your final product. I said you'd be better off to use a standalone DVD recorder in your situation. >>
I'm not trying to say anything back to you, I was just letting you know the process I'm using. With any stand alone dvd-recorder I can't manipulate the final DVD product with 5% of the functionality of Senarist, and that is functionality I need. What I'm doing isn't just a project for myself. It's more of a small business kind of thing.

My sources are mainly NTSC VHS tape for now, but in the future.. who knows.

The whole point getting this deck is to get the highest possible quality from the VHS tape, (as well as pretty much any other source) so from there it's all digital, where my final product can be up to very strict specifications. Professional I guess.

[FredThompson]

You still don't understand.

NTSC DV is 4:1:1, got it? It's NOT 4:2:0. NTSC DV is horrible for thin lines and any CGI stuff.

NTSC DVD is 4:2:0. If you go from 4:1:1 to 4:2:0 you get only half the chroma resolution which is possible.

What might you be able to do with a 15M datarate DVD produced from a standalone DVD recorder? Perhaps...put it in a computer and read the video files from it?

You will not get good results from ANY NTSC DV device. DV is convenient but it is horrible for details of any sort.

As your final end product is to be 4:2:0, you should have at least the resolution in every step along the way. If you use DV, you'll royally screw up the chroma, capice?

Find the thread over in one fo the AviSynth areas where Xesdeeni and I talk about the problems of DV . There are screenshots which will help you understand.

[2ZOD.COM]

I see, now it's all making more sense.

Either way, this deck is still pretty much exactly what I'm looking for. Tons of features.

I'm looking for that thread you're talking about, but I'm yet to find it.

[FredThompson]

That deck's features won't prevent the problem of NTSC DV.

I can't remember the name of the thread. I asked about a faster lanczos routine than the AviSynth internal version. The title is something like that.

[2ZOD.COM]

<< That descks features won't prevent the problem of NTSC DV. >>
I can still use the S-Video out. I would assume that it converts to DV only before it outputs it through firewire.

Damn, i remember skimming past that thread. I read a lot of it, but didn't see where you guys started to discuss DV. Blasted..

[2ZOD.COM]

<< NTSC DV is 4:1:1, got it? It's NOT 4:2:0. NTSC DV is horrible for thin lines and any CGI stuff. >>
Wait wait wiat... NTSC VHS is 4:1:1 too isn't it? It wouldn't do anything for the quality (towards the negative side) converting VHS tapes to DV. If anything it would still boost the quality due to the fact that you don't have any other a/d conversions. If you use the S-Video out there is another (much worse quality) a/d conversion in the capture card.

[FredThompson]

You are absolutely incorrect. While NTSC VHS has certain performance characteristics, you cannot somehow losslessly translate it to DV.

Borrow some equipment and compare DV capture of VHS to MPEG2 capture of VHS. You'll find MPEG2 is far better for details.

The last part of your comment makes no sense. Are you proposing to somehow hack into internal frame cleanup electronics in a good VCR?

[FredThompson]

OK, go here: http://www.geocities.com/xesdeeni2001/

Look at the 4:1:1 comparison. The MainConcept one shows what you will get with DV. Original is from a FlyVideo capture.

The source is the Star Wars laserdisc. All laserdiscs are composite so that's the best way to capture them. However, with VHS tape, you're going to get better results using S-Video out, not composite.

What I'm proposing is that given your final goal is MPEG2, you don't need to capture full lossless, you need a good MPEG2 capture to start. If you capture at 15M/sec you'll be far above your final destination and you won't have to change color models anywhere in the processing chain.

You CAN capture lossless but why do that if you're going to MPEG2? It will have to be downsampled anyhow. Sure, it looks a little nicer on the screen while you edit but all that will be lost in the end.

Make sense?

[2ZOD.COM]

<< The last part of your comment makes no sense. Are you proposing to somehow hack into internal frame cleanup electronics in a good VCR? >>

Sorry, I messed up. I meant to say S-Video, not X-Video. I edited it. I was in a bit of a hurry.

You said you used DV Pass-Through when you capture from VHS via S-Video, correct? Doesn't that mean there's on-the-fly DV encoding taking place within the DV pass-through device? If that's the case you're completely bashing what you said I wanted to do. I'm sorry, I must not understand the dv pass-through process enough. Could you give me the basics of it?

MPEG-2 15mbit or Lossless, either way it has to be re-encoded for DVD. So either way I'm stuck encoding. I know that simply lowering the bitrate for the MPEG-2 is significantly faster, but the quality overall will be lower due to the original MPEG-2 capture.

<< Sure, it looks a little nicer on the screen while you edit but all that will be lost in the end. >>

I don't understand how it would be lost? You see so much more detail on a computer monitor than on the TV, and I'm sure most of the fine details go completely unnoticed on a TV, but that doesn't mean they shouldn't be there. I'm not sure if that's what you're refering to when you say they'll be lost, or you may be refering to the down sampling with lossless.

<< It will have to be downsampled anyhow. >>

Bottom of the line here is Lossless compression converted to MPEG-2, that isn't done in real-time, will produce superior results to any on-the-fly MPEG-2 encoding hardware. (that is within my price range at least) Overall it would be a matter of processing time. Is worth the slight added quality when you take the extra time to convert Lossless to MPEG-2? I say yes.

Please tell me if I'm totally wrong here.

[FredThompson]

OK, I think I understand now.

Yes, I use either DV passthru or an extermal MPEG2 encoder. I far prefer to use the encoder because the end result is better. That doesn't mean I'm throwing away the DV camcorder so the capability is here, I just don't use it much anymore.

I to use the digital filtering and stabilization of my JVC HR-S7600U and an S-Video connection. There will usually be a few lines at the bottom which need to be cropped and sometimes things skew at the top. However, as I'm going to watch on a TV, I'll typically overlay those areas with black bars. The circuitry in this VCR makes black black, not a swirl of red and blue. It also stabilizes the image. Admittedly, this does change the signal somewhat but the trade-off is worth it. If you've never seen this type of filtering, go to a good A/V store and have them demonstrate it for you. The best way to remove this type of noise and problem is before it goes out the video cables. Processing the captured signals can only do so much. GIGO.

I only use composite for PAL because I don't have an S-VHS PAL deck. Laserdiscs are stored as composite signals and that's the proper way to play them.

NTSC DV is 4:1:1 which means one color value for 4 adjacent horizontal pixels. MPEG2 is 4:2:0 which means one color value for a 2 x 2 block of 4 pixels. Thus, DV has double the vertical color resolution and half the horizontal resolution of MPEG2. Re-read that if you aren't sure you understand it.

OK, When you have a hard edge, thin vertical line, or slight diagonal it is better represented visually by the 4:2:0 "cube" model than the 4:1:1 "sliver" model. DV yields a lot of shimmering on video bugs (the little station thing you see in the lower right corner), thin lines and metal. I have some DV I took in a factory of a machine in motion which is almost unbearable to watch because the bars look like a series of chunks, not smooth lines.

In any event, your end result going to MPEG2 is 4:2:0, right? So, your will have a final product in which the color is saved as one color value for a 2 x 2 block of 4 pixels. That is one of the ways the size compression happens, you only keep 1/4 of "full" color information.

Given that, lossless capture and all the processing you do on a computer will look wonderful on the computer but you'll LOSE 3/4 of that color information during the process of encoding to MPEG2.

If you convert NTSC DV's 4:1:1 to DVD's MPEG2 4:2:0 you will only have half the horizontal color information which MPEG2 4:2:0 can support. You do have double the vertical color resolution but that gets converted. The result is a 4:2:0 color model in which the horizontal aspect of the color is roughly half as good as it could be.

This is an imperfect analogy but think of the difference between a plaid shirt and a checkered shirt. Each square of the checkered shirt has a color which is not dependent upon the adjacent square. The plaid shirt isn't like that. Converting NTSC DV 4:1:1 to DVD MPEG2 4:2:0 is kind of like the plaid shirt. Your horizontal color information is going to be distorted.

That's why I don't like to use NTSC DV. Unfortunately, there isn't much of an option for camcorder use so I'm stuck with it. PAL DV is 4:2:0 but isn't fully compatible so...you get the idea.

OK, so now we've determined that if the final product is to be MPEG2 for a DVD, we really DON'T want to use NTSC DV as the source if we can avoid it.

What we want is a method of capture such that we have at least the MPEG2 4:2:0 color information. IOW, it would be better to throw away color information as part of the encoding process than to be missing color information. Make sense?

There are really only 2 options; full lossless capture or 4:2:0 capture. (You could do 4:2:2, etc. but the equipment gets more exotic and expensive. Granted, the end result will be somewhat better.)

What I am proposing is that given your desired DVD MPEG2 4:2:0 output, you can capture as 4:2:0 with a significantly higher bitrate, do your filtering and processing with 4:2:0-aware methods, and have very nice DVD MPEG2 4:2:0 results.

Additionally, you will save processing time and storage space.

FWIW, one of the "names" you will come to recognize as you hang out in the AviSynth/VirtualDub arena frequently uses a standalone DVD recorder at the highest bitrate instead of lossless capture because teh combination of convenience and quality is pretty good.

The dual tape deck you mentioned is nice but quite expensive. It looks to be a couple of hundred dollars more than buying a basic DV camcorder and a good S-VHS deck.

[2ZOD.COM]

OK, now I understand the colorspace part. I never fully understood it as it's relation to actual video quality. Thank you for that little lesson.

For some reason I didn't remember that your post before also said you use the MPEG-2 capture device.

The whole idea of a VCR with firewire out as I said much earlier in this topic was more of a high quality VCR, one with features like the JVC SR-VS30U, but the firewire out would actually be more of an uncompressed kind of thing. I know such things exist, and are strictly broadcast quality output in RGB directly, with INSANE data rates, but I mean more of something with another format (maybe even proprietary) with a better colorspace. Do you get what I mean. When I said firewire, I didn't exactly imply that it was to be NTSC DV format. The main concept behind my idea was that the source tape would just be converted to a digital source as close to the origin as possible. This would be the most practical for the home/pro-sumer market, where we can't afford a Ultra320 SCSI array to handle insane bandwidths of straight RGB. I just assumed the 400mbit firewire bandwidth would be sufficient for most needs.


My whole concept is to get as much quality as possible. Now I understand that using the DV Passthru method, as well as if I were to use that JVC vcr witht he DV out via firewire I would actually get reduced overall quality.

I still stand firm on my lossless to mpeg-2 method. Even if it isn't with the DV method, S-Video would still most defineatly be sufficient. The processing time is not even an issue.

I really do have to thank you for providing so much insight on capturing video here. I spent a good 2 hours yesterday trying to find something about different colorspaces all to no real avail. Your few-sentance explaination did it better than anything. Thanks.

[FredThompson]

If I were in your situation, I'd buy a good capture card to lossless capture, a good high-end S-VHS deck and some hefty, fast drives. Make sure you understand just how much space you're going to need. With uncompressed captures, huge amounts of time are involved in just moving through the files.

I'll be interested to hear you opinion after 6-8 months of 2-hour uncompressed captures.

Thanks for the kind words. I tried to explain as clearly as I now understand the real-world effects of the color models.

[2ZOD.COM]

For the past few months I've been using a Fusion 878A based card with HuffyUV. Not uncompressed, but lossless and still large files sizes. I also have about 350gb of free hard drive space dedicated to the video capture and authoring. I've been doing short video conversions (5-45 min) for a little bit now, but I really think I should get some pro-sumer hardware. That's why I started this thread in the beginning.

When I was looking at those decks you were talking about I found another model that has S-Video out and is 1/3 of the price. Since what I'm really looking for doesn't exactly exist, and all I really need, I'm probably going for that; exactly what you said. Then some highre quality capture hardware like Pinnacle or something. I'm not sure about the capture device exactly, but I'm defineatly getting the S-VHS vcr.

[jfcarbel]

>> Laserdiscs are stored as composite signals and that's the proper way to play them.

Fred, I have an S-VHS out on my Laserdisc Player and I thought that gives better quality and higher resolution, how then is it better to capture using the older composite?

I have a Matrox RTX.100 and have been doing all my captures in DV from a JVC VHS S-video input and then compressing to MPEG2 using VBR with average bitrate 5500 and 3 pass. And the quality looks great. But Matrox does have a realtime MPEG2 capture feature, would you say this would give me better quality?

I done many captures from DirecTV s-video to DV and then to MPEG2 encoder and the quality seems great, is it just my eyes are not that good? Or maybe its because most of my captures do not include tough scenes like action and CGI.

I am asking about LD because my next project is converting my Star Wars trilogy LD to DVD, since Lucas is not release the original versions.

[FredThompson]

Laserdiscs are analog storage of composite signals. You can compare captures from the composite and s-video out but the composite will be better. I know it seems odd because S-video is better for tape playback. I don't understand the "why" completely, I just know this is the case. Maybe it has something to do with optical reading vs. magnetic reading.

Be sure to use Reinterpolate411 or 411Helper with your NTSC DV captures.

IanD has been doing a lot of tests trying to get the ultimate capture from the Star Wars laserdiscs. There are a few posts from him here at doom9. The screen captures the are in the discussions of Reinterpolate411 and 411Helper were made by him from the opening scene. NTSC DV is horrible for saturated color on a black background.

[jfcarbel]

What about this does it allow capture or is it only for decoding? It uses the 4:2:2 colorspace. What does Huffy codec use and does it make since to maybe use it and then go to MPEG-2?

The MainConcept DVCPro 25/50 DV Codec is our new product for our amateur and professional customers working with high end Video. DVCPRO50 delivers considerably higher picture quality since it uses less compression than DV. The Codec comes with a full-featured DirectShow filter. It is compatible with any program which supports the standard DirectShow interface.

Technical Specifications:

DVCPro 50: 4:2:2 Color Space, 8 Bit Quantization, data reduction of 5:1, data rate of 50 MBit/s

[Arachnotron]

Laserdiscs are analog storage of composite signals. You can compare captures from the composite and s-video out but the composite will be better. I know it seems odd because S-video is better for tape playback. I don't understand the "why" completely, I just know this is the case.

Is it not simply a case of which device has the better comb filter?
The signal stored on the disk is composite, so for s-video out the player has to combfilter it first. Laserdisk players, being an older technology, probably contain more primitive comb filters then modern capture devices do, so it is better to use the composite signal and leave the separation to the newer device.

VHS tape on the other hand, stores Y and C more or less separate. So to create a composite signal, Y and C are joined again on playback, after which the cap device has to separate them again. Comb filtering will allways degrade the picture somewhat. Using a VCR with s-vodeo out skips an extra comb filtering step.


What does Huffy codec use

yuv2 I think, so 4:2:2.

[vhelp]

hi guys,

I was checking Freds (non-commercial) website (thanks) and D/L'ed
the many sample pics and .avi's avail. I saw Star Wars .avi clip and
couldn't help to wonder how well it looked, as a VHS capture. (my
confusion due to the Flywheel card being discussed) anyways..
I skimed it w/ vdub and noticed how close it look to my VHS capture
(s-video) and widescreen version. I was amazed how the pixles were
so closely resembling each other. But, the clarity and uniformity
was much greater than my VHS captured version. I said, "what VCR did
this guy use, hehe" then I realized, it was LD - boing.
.
.
But, the colors on that LD was more richer. Was that from a PAL disk ?
From my dealing w/ PAL sources, they are obviously better than the
counter part, NTSC. Most definately.

Laserdisk players, being an older technology, probably contain more primitive comb filters then modern capture devices do, so it is better to use the composite signal and leave the separation to the newer device.

not comparing dvd, I would say that LD is way better then VHS :) even
for it's mentioned, "premitive comb filter".

@ Archnotron,

The noise level on that LD (above) was, if at all, very low, to faint.
Nothing like my VHS version. I'v will U/L'ed a version for you to compare
and discuss (if you feel the need or aid in this thread) :) here (http://www.westnet.com/~vhelp/vhelp.sw(se.ws).openscene.vhs.jvc(hr_s3910u).svideo.advc.avi)
Note, this is an untouched/unfilted DV version.

I noticed that there were more stars in my vhs version. I'm wondering
if either, there were more SE versions made, or was it due to the
nature of VHS and color loss or detail. I'm sure it was not noise being
added. Notice the lower center, where you see a patch of stars in the vhs
version vs. the LD versions. But, to make a fair assessment, one has to
have both the SE Widescreen version on VHS and LD to analise with.
Just to note, I'm using Fred's "SWdv.avi" sample from his website.

@ fred, feel free to add a copy of this NTSC version to your website, if it
will help w/ your research, and name it what you want, if need be.

As to the capture side of VHS (at least in mine) and my ADVC, I feel it is
as ~professional (from a homeowner POV, as can ben) due to the capabilities of
the DV device itself. About the only issue I have w/ DV, is the colorspace
for NTSC. It's lower. And, trying to pump it up will not help, unless
the process were done in a "digitized" mannor, of which we do not have.
Turning on IRE=7.5 for NTSC (though I used to do this in vain) only make
it worse, because the color become uncontrolable.. if that makes any sense.
I found that leaving the NTSC version of the advc's IRE=0 was the best
setting to use, especially for VHS projects.. and, no color adjustments
should be made - ever. In my opinion, leaving this setting alone and not
tweaking the colors (ie, under TMPG) will give you precise quality, especially
when using low bitrates in your projects (especially for VHS sources)
Setting the IRE=7.5 for NTSC for your Satellite projects is another topic.
but, in short, it's not as bad, and can yield good results at best.

From a limited homeowner budget, the advc is as professional as you'll
ever get. However, I still feel that an Analog capture device is THE WAY
to go, becuase of the greater color space it provides, and the other
possibilities (ie, codecs; or non-codec; resolutions; etc) but due to the
great numbers w/ users having audio sync (drift) issues, the device I
mention is usulally the end result. It was for me :)

I hope I've ben of some help here,
-vhelp

[Arachnotron]

I would say that LD is way better then VHS even

I never disputed that. I was just explaining why with VHS you often get better results using a s-video connection to the cap device, while with laserdisk the composite out gives better result.

The bandwidth available to laserdisk is higher than VHS tape, so laserdisk will give you more detail, especially with color.

But the latest cap devices have superior combfiltering. I believe the ADVC300 even has a frame buffer and can do 3D combfiltering. Feeding it the composite signal from a LD player should give superior quality compared to using the s-video out, unless you happen to have an LD player with a better combfilter than that.

I noticed that there were more stars in my vhs version. I'm wondering
if either, there were more SE versions made

I thought the latest VHS release contained the version with added digital effects, so there might very well be more stars in space scenes on it.

Was this new version even released on laserdisk? You should be able to see the difference in scenes with jabba the Hutt in it; AFAIKR the first version was a puppet, in the second release he was a computer animation.

[Abaddon666]

Hi,

I read all the posts in this thread and I hope you
guys can help me with my problem. Well first of all,
my conversions are analog Pal->dv Pal->DVD, so I won't
have the 4:1:1 problem, since DV-Pal is 4:2:0, right ?

I use the ADVC-100 by Canopus, but since I have cable-TV
and an old VHS-VCR as source, my results look quite bad.

I plan to get the JVC-HR-S8960E (I live in Germany, so the
model might have another name in the US).
It has this so-called "digiPure"-Feature (DNR, digital TBC,...).

Now, what I would like to know is:
Do these SVHS-Decks only improve video-quality during playback
from tape or are they even capable of improving the cable-tv-input
and passing it out through S-Video ?
I'm looking for a solution for receiving good signals in both
cases, playback from tape AND TV-Broadcasting.

Even if this wasn't the case, I'd still have some advantage of
using this SVHS-Deck I guess, since it will at least put out a
S-Video signal (my old VHS, which I use as source for tapes and
tuner only puts out composite). But if it really can't clean
the cable-TV input in the way I described above, then I might need
to get another solution.

I would appreciate any kind of advice, regarding this.

Tanks in advance,


Abaddon666

[FredThompson]

PAL does not have the 4:1:1->4:2:0 conversion issues NTSC does.

AFAIK, the cleaning is only for tape playback.