Dropping 100s of frames!!! Please help!!!

[proxyx99]

Jesus...!

Now I don't know whether to laugh or cry?

.....
Clanger's link to a review of my capture card was a bit of an eye-opener. Maybe that's my problem??

Sheesh...who knows!?!?!?

I'm havin' a beer.

I'm afraid the the issue ends here. In my first post I've suggested calling Leadtek in order to find out what you can reasonably expect from this product... What Clanger did, is taken it a step further to find an answer for you. Reading the linked material confirmed my suspitions that maybe the expectations are set a bit too high. If not for that... You need to find out by learning directly from horses mouth (Leadtek) that this is not the case.

[rhegedus]

I think the problem lies with having just one hard drive. No matter how you partition it, it will still be the same drive that you're OS is on. Every time you send/receive mail, surf the net, print a page, whatever, you will still drop frames.

That's even before we get onto other stuff like capture format (compressed/uncompressed avi, mpeg etc), capture cards and sound cards, clock cycles etc.

Sort the simplest stuff out first.

[johns0]

How about trying a different capture card?Not the same brand,that will tell you whether its the capture card or your computer thats at fault.

[proxyx99]

Quick comment. If you have HD partitioned in several volumes just use Sandra or any decent HD benchmark and test it.
You'll find out that empty vol. may be up to 30% faster (!). Best combination: outside HD track (fastest) plus no files on it. The only way to find out is to do the test yourself. I would encourage you to do it. Good way to find out on which volume to put your pagefile...

[lordsmurf]

Thanks, guys. This forum rocks!

Ok - firstly yep I disable virus protection before capturing! (--and screensavers, firewalls, etc etc)...

Hmm...resolutions...well, I figure I have to go the whole hog (720x576) in order to achieve good results on DVD!? Capturing at a lower resolution is swell for VCD - and I've achieved excellent results!!! - but to capture at a lower res, then burn to DVD...well, the quality sux!!! Yes, yes, I know: garbage in, garbage out, and all that. That's not really the issue here. The issue is, I can capture at 352x288 (for PAL) and burn to VCD, with great results, as good if not better than the VHS source!!! But if, say, I capture at 384x576, or 384x288, or whatever, and then burn to DVD, the quality is shocking, real blurry & heaps worse than the VHS source!!! I figure I have to capture at a high res (eg 720x576) in order to achieve satisfactory transfer to DVD!? I know I can capture my VHS tapes with great quality at a lower resolution, and it looks great on my monitor, but then the problem becomes burning it to DVD - all that great quality disappears!

Or am I barking up the wrong tree here????

Anyhow, there's a few suggestions here I will try. OF COURSE I will post my results!!!!

Thanks again for the responses!!!!!!!!!!!!!!!

Yes, you are FAR, FAR off base.
This page was meant for people like yourself:
http://www.digitalfaq.com/dvdguides/capture/understandsource.htm
Read if from top to bottom. Spend 5-10 minutes or whatever it takes.

[trevlac]

This page was meant for people like yourself:
http://www.digitalfaq.com/dvdguides/capture/understandsource.htm
Read if from top to bottom. Spend 5-10 minutes or whatever it takes.

Yes! very good summary. Well written. I read it a while ago, and just now again. Worth the few minutes. Too bad we can't make comments on it. Can external guides be commented on? Only thing I don't like is the Broadcast TV is "336x480 in digital terms" thing. VHS is most definately sub 352. Also, you could add some sources

Problem is there is too much info out there for PC watchin divx encoding freaky people. 384x288(576) are PC watchin numbers.

[Co0L]

Just another comment for this issue!
1. Verify that You are using latest WDM drivers.
2. Update graphics drivers
3. Update DirectX (Yes, it really matters!)
4. Turn off ALL unnecessary 'sh*t' on Your PC.
(meaning AV progs and similar)
5. Always start with defragged HD.

Finally, capturing from VHS to PC is a nightmare...

Nuff said, and good luck!

[2man]

i only get dropped frames when the tape isn´t good
and when i´m capturing w/frame rate different than 25 fps(my system colour is pal)
so mateo check the frame rate on rigth-bottom if you are capturing 25fps
and try uncompressed avi

[Nelson37]

Trevlac - I am not angry with you, or anyone else for that matter, but this insistence on using math, formulas, charts, or in fact anything at all other than LOOKING AT IT to "prove" that one capture resolution is "best" for a given source really annoys me. It has been repeated on this board many times, and it just totally ignores so many important factors.

Data IS important, such as the recently learned info that all cards cap at a certain res and then resize to give you what you asked for. This tidbit in fact probably explains my observed results, and also bolsters my point.
The difference of 704 over 720 is slight, although beneficial. This res is not currently available to me, though the new MMC 8.8 apparently fixes this. Capturing cropped seems to mitigate the error, though.

[trevlac]

Trevlac - I am not angry with you, or anyone else for that matter, but this insistence on using math, formulas, charts, or in fact anything at all other than LOOKING AT IT to "prove" that one capture resolution is "best" for a given source really annoys me. It has been repeated on this board many times, and it just totally ignores so many important factors.

I'm glad your not angry. Sometimes I misconstru CAPS AS YELLING. Really. Smurf just pointed out to me that he does not mean that when he uses caps.

On the math/theory vs observation stuff.... don't you want to know why one looks better than the other? And if the theory does not match the practice, why that is too ?

And yes, for a given setup and source, the best way to know is to test. That's what I do. However, a reasonable understanding of the theory should help you know what to test and what to tweak. Sometimes there are so many options.

BTW: I am really just learning this stuff. As a hobby, I am very interested in the inner workings. In fact. I hardly watch any movies/tv at all. I may have completely different goals than others around here.

Cheers

[Matteo693]

OK, I've just noticed something. When I'm capturing with Virtual Dub (using huffyuv), I notice the data rate creeps up 6000 to 7000 to 8000 to 9000, at which point frames start dropping like crazy & the data rate starts declining. PicVideo MJPEG works (at full resolution) - data rate stays around 3000-4000. Could this be the problem? How do I stabilize the data rate under 8000Mb/s using huffyuv??

Sorry for the run-around, but yeah I AM just a newbie at this... Anyway, I'm sure there's tons of helpful hints in this post for other folk out there...!?

NB: Yep, all my WDM & chipset drivers are updated.

Lordsmurf, yeah I'm afraid I was resizing a lower-res capture up to 720x576 for DVD, that was why the image was blurring out. My fault.

Still, the point of my original posting was why can't I capture at full res?

The capture guide over at doom9.org clearly differentiates between "High Quality" VHS capture (720x576), "Average Quality" capture (384x576) and "Poor Quality" capture (384x240)...(might have those res's confused, but you get the gist)...Anyway, this seems to go against what everybody here is saying? I don't get it?

Well...over to you...!

[FOO]

Regarding Horizontal sampling rate...

On the math/theory vs observation stuff.... don't you want to know why one looks better than the other? And if the theory does not match the practice, why that is too ?

People have been using the Nyquist criterion here to prove that a certain sampling rate is adequate. ..or rather that more is useless.
It is also claimed that 720 LOOKs better that 352, apparenly at odds with
theory. The problem here is this is not an ideal situation , and the theory
does apply to it.

Theoretically , sampling at twice the highest frequency present in the
source is sufficient to reproduce it exactly. However this is true ONLY
if a PERFECT low pass filter is applied to the samples to reconstruct them.
Perfect low pass filters are a physical impossibility , they don't cut
frequencies immediately above the cutoff to zero , they gradually
"roll" off. This requires that in real equipment , the sample rate is
set a little higher than 2f , depending on how tight the LPF is.

44100 should be able to grab audio to 22050 Hz right ? So why
did DVD go to 48000 when people "cutoff" at 20000. It's to give
the filter some slack.

I see some difference capping tape at 720 vs 352. I think that 640
would be OK , but if you do 720 you don't have to resize

[Nelson37]

Matteo - incrementally lowering the resolution, and/or eliminating audio, is the only way I know of to controllably lower the filesize, or datarate, using a Huffy capture. This will find the threshold at which your Hard Drive will be able to continously write data, which appears to be your problem. A second hard drive on a seperate channel is the most tried-and-true way to solve this problem. Using MJPEG is another way with a slight loss in quality, though you did not mention what setting you used? I believe set at 20 files are only slightly smaller than Huffy, 18 or 19 is about half the size, etc. That data writing threshold is what you need to find. Then you need to either operate within it or increase it somehow.

That is why I previously suggested all these steps. That is why they are suggested in the guides and stickys about lowering frame drops.

Although someone is certain to tell you that your drive is theoretically capable on maintaining the data rate you need, in contrast to your real-world experience.

Examining the difference between the two may very well lead to useful info, or just fascinating tidbits of trivia. It may also tell you that the theory just doesn't apply in your case. Or the theory may in fact be accurate but is being affected by some other factor. Such as the fact that some of the available data rate is being used for system tasks other than your capture. Or that your video card cannot actually cap at 352x but uses an imperfect resize to acheive that res.

Anybody remember when theory "proved" that the bumblebee could not fly? The problem was that they had incorrectly measured the wing speed, and had not properly compensated for the changes in angle of attack.

I am also reminded of the ancient Greeks who debated how many teeth there were in a horse's mouth without ever actually examining a horse.

[FOO]

It might be more useful to know hown many teeth there are
in a horse's ass

[trevlac]

Regarding Horizontal sampling rate...

Theoretically , sampling at twice the highest frequency present in the
source is sufficient to reproduce it exactly. However this is true ONLY
if a PERFECT low pass filter is applied to the samples to reconstruct them.
Perfect low pass filters are a physical impossibility , they don't cut
frequencies immediately above the cutoff to zero , they gradually
"roll" off. This requires that in real equipment , the sample rate is
set a little higher than 2f , depending on how tight the LPF is.

My understanding is close to this, but not exactly. 2f is sufficient to create the wave exactly. The lowpass filter is intended to remove frequency higher than 1/2f. If you apply NO filter, you can't be sure all frequencies are 1/2f, and therefore, you risk ailasing (or high frequencies blending with lower frequencies). AKA noise.

44100 should be able to grab audio to 22050 Hz right ? So why
did DVD go to 48000 when people "cutoff" at 20000. It's to give
the filter some slack.

Audio is not video. Portions of an audio wave's phase can be changed and it would sound the same. Frequency is the important thing in an audio wave. Shape is the important thing in a video wave. Bottom line, frequency aliasing is a BAD thing for audio. Not for video. That is why they super sample. I'd bet Professional video equipment does not sample higher than 27MHz. Actually, a low pass filter can cause more damage to the shape of a wave.

I see some difference capping tape at 720 vs 352. I think that 640
would be OK , but if you do 720 you don't have to resize

640 is an interesting number. That is about the limit of what a BT878 can do.


@Nelson37

As far as not looking at the horse: We all can read the spec of a BT878a. 172 pages worth.


Here is what it says for NTSC. (As far as I can tell ) The chip samples at 28.636MHz or 8x the color frequency of 3.58MHz. There is no filter before sample. Then a digital lowpass filter is applied, killing all frequencies above 6MHz. (I'd have to know Fourier Transforms to explain this filter, and you'd have to know them to understand.) Next step, throw out 1/2 the samples. This effectively puts you at a sample rate of 14.318. Keep in mind that all of this takes place regardless of the size you requested.

Now, if the digital filter works, you get zero resolution boost by asking for anything over ~640. (There is of course the spec size bennifit, but as i mentioned before, 720 is not the right size.)

Finally, if you ask for anything less than 640, you risk digital noise (aliasing), unless an additional lowpass filter is applied. This risk really depends upon your source. VHS has no additional resolution above 3MHz, however, analog noise between 3MHz and 6MHz could cause aliasing.

I would think this risk depends upon you signal, so a test would be in order.

For the BTwincap driver. When you hit < 3.5MHz or <360 width, the driver applies a 3MHz filter. The chip supports up to a 5-tap filter. The driver i mentioned sets a 2-tap option. You've got me as to what this means.

Maybe based upon this, you can decide what you want to do.



@Matteo693

Doom9's guide is quite good. However, I think if you pressed Wilbert, he'd admit there are a few flaws. One is clearly the statement to cap at as high as possible. 640 is the limit for a BT878 card. I doubt any normal analog source reaches this limit. So what's the point of doing 768x576 ?
Maybe I'll post over there and give him hell.

By the way, cutting the vertical res in 1/2 would kill the resolution. Even vhs has full vertical resolution. Talk about low quality. Talk about VCD.


I appoligize ahead of time if I seem to be debating just to debate. Not true. I find if I know what is happening inside of the black box, I can make better decisions. You people seem to have a really good understanding of what's happening. I'm just trying to improve mine. (short of taking TV engineering courses)


Trev

edit If anyone cares to try to go beyond black box testing, here is a link that lists a few data sheets for some Conexant and Philips chips. You'd have to have some idea of theory to understand them. FOO clearly does. Nelson, I'm not sure about you. All theory is bad? :P

http://yanyan.dtdns.net/capture/

[FOO]

My understanding is close to this, but not exactly. 2f is sufficient to create the wave exactly. The lowpass filter is intended to remove frequency higher than 1/2f. If you apply NO filter, you can't be sure all frequencies are 1/2f, and therefore, you risk ailasing (or high frequencies blending with lower frequencies). AKA noise.

Nah

The ACT of sampling creates multiple copies of the spectrum of the
sampled waveform up the frequency axis with a spacing at the sampling frequency. It's not a "risk" problem. It's guaranteed. You MUST low pass filter the sampled spectrum at 1/2 sampling rate or you get garbage.

Bottom line, frequency aliasing is a BAD thing for audio. Not for video.

Aliasing trashes the signal . Period . It is not ok for video.

[lordsmurf]

The capture guide over at doom9.org clearly differentiates between "High Quality" VHS capture (720x576), "Average Quality" capture (384x576) and "Poor Quality" capture (384x240)...(might have those res's confused, but you get the gist)...Anyway, this seems to go against what everybody here is saying? I don't get it?

When it comes to capturing, I find most doom9 information to be off the mark. They are best left to information about converting stuff and ripping DVD, not capturing. Their capture forum is actually quite puny.

[trevlac]

Nah

The ACT of sampling creates multiple copies of the spectrum of the
sampled waveform up the frequency axis with a spacing at the sampling frequency.

Description of sampling in the frequency domain. Not aliasing.

There are many ways for information to be encoded in an analog waveform, two are common, time domain (video), and frequency domain (audio). Since aliasing misplaces and overlaps frequency components, it directly destroys information encoded in the frequency domain. In the time domain, Using no filter says the value of each sample is identical to the value of the orignial signal. In other words, it has the perfect edge sharpness. However, you also get noise. The amount depends on the signal. No guarentees that it is trashed.

It's not a "risk" problem. It's guaranteed. You MUST low pass filter the sampled spectrum at 1/2 sampling rate or you get garbage.

Well the BT878 spec clearly states that they do not filter before they sample. They filter after. They do assume that a normal video signal will have little noise above 28MHz. If there is a guarantee that anything above 28MHz trashes the picture, then they have a problem.

As another note. If you resize from 640 to 480, you are effectivly droping your sample rate. Sample at 27MHz and throw out everyother sample gives you 13.5MHz. Does this trash the picture ?

You of course may think the 'spec' is false. I can't say, I'm just a Banker. Not an EE.

[trevlac]

When it comes to capturing, I find most doom9 information to be off the mark. They are best left to information about converting stuff and ripping DVD, not capturing. Their capture forum is actually quite puny.

Since I said wilbert's guide was not 100%, I felt I had to correct this sillyness.

To say it nicely, LS has a completely different point of view than the doom9ers. For example, he preferrs 'video soap'. They like and write avisynth filters.

Then again, a bunch of them live for anime. Smurf likes, well, smuffs.

Actually, I'm with LS on this last point. 'cept, I tend to like old hanna barbara

[FOO]

There are many ways for information to be encoded in an analog waveform, two are common, time domain (video), and frequency domain (audio). Since aliasing misplaces and overlaps frequency components, it directly destroys information encoded in the frequency domain. In the time domain, Using no filter says the value of each sample is identical to the value of the orignial signal. In other words, it has the perfect edge sharpness. However, you also get noise. The amount depends on the signal. No guarentees that it is trashed.

Most of this is not true. I guess you don't believe that sampling changes the spectrum. The fact that people are insensitive to audio phase does
not mean audio is represented in the frequency domain either

I guess I'll have to go find a picture

[trevlac]

Most of this is not true. I guess you don't believe that sampling changes the spectrum.

Foo,

Sampling does change the wave. Can't say what I believe. I'm still learning. However, I'm just paraphrasing sources. I'm not making it up. See this one about page 45 to the end.

The Scientist and Engineer's Guide to Digital Signal Processing Chapter 3 (ADC and DAC) Also available on amazon.

This one has a great section on sampling: Snell & Wilcox - Engineers Guide to Compression

This one has some nice pictures with regard to resizing samples.
http://www.snellwilcox.com/knowledgecenter/whitepapers/papers/digital_arc.pdf

Here is another good one: Basic signal Processing

Now I may be misreading these things, but I am trying to learn.

How about this, is it true? Assume when you ask for a size of say 368 the chip discards samples. Now, isn't doing that effectively reducing the sampling rate? And doing that should introduce aliasing? Finally, does that "trash" the picture?

Foo, I did want to say, I am greatful for your patience. Free education is wonderful.



@Lordsmurf

Before you dismiss Doom9. I got those links, and the BT878 spec link there. Mainly from the capture guide list of references. Real nice when a guide provides references.....

[FOO]

OK found picture.
1. Original signal with spectrum
2. Sampled signal with spectrum . Note extra crap
3. Low pass filter with too high cutoff. Note signal is garbage
4. same as #2
5. Correct low pass filter . Signal now Happy

[trevlac]

FOO,

I like your picture, but seems we are talking about 2 different things here.

Your picture describes a problem with Digital to Analog conversion. A problem for your DVD player. I'm talking about Analog to Digital conversion. A challange for your capture card.

This is my take on your picture:

1 - The analog Wave
2 - Sample it. Now we are in the digital realm
3 - Filter out the samples to get back to analog (DAC) Filter was not 'tight' enough around the sample frequency.
4 - Same as 2
5 - Good Digital to Analog Conversion.

How 'bout ADC ?

[FOO]

The only reason for the picture is that I thought I heard you say
that a LPF was not necessary or was optional.

I guess you are trying to figure out what the horizontal sampling rate
should be for a VHS tape capture . Is this true ?

I regularly see statements claiming that sampling higher than the cutoff
of a VHS tape or a Tuner are useless, and it's obvious to me it's not so
just by looking. I was also trying to address some of that.
There is more to it than just Nyquist.
Oversampling helps even when it's theoretically not required.

Before I start a war , I'll say that any improvement with oversampling is
minimal.

[trevlac]

FOO,

I appreciate and respect your comments. You make me think, and I like it. You are also trying to real us back (close) to the topic. Thanks. I'm also not really interested in fighting about it. Ironnically, I think we have the same conclusion: "It all depends on your source and equipement." Test and you'll know.

I did want to address your concens, and try to summarize my position. But I'm not sure there is much more for me to say (unless I take those courses )

The only reason for the picture is that I thought I heard you say
that a LPF was not necessary or was optional.

LPF is optional before sampling a video wave. That is what the BT878 does.

I guess you are trying to figure out what the horizontal sampling rate
should be for a VHS tape capture . Is this true ?

Just trying to have some guide. From this, one needs to know ones hardware.

I regularly see statements claiming that sampling higher than the cutoff
of a VHS tape or a Tuner are useless, and it's obvious to me it's not so
just by looking. I was also trying to address some of that.

Well the theory is sound here. The implementation is in question. Also, strangely enough, Lordsmurf repeats that from experience, he sees zero difference between 352 and higher for VHS. The reason this is strange is because he, like you, uses ATI equipment.

There is more to it than just Nyquist.
Oversampling helps even when it's theoretically not required.

There is more to ADC than just Nyquist. However, oversampling helps if you filter. Get's you nothing if you don't. PS: Good filter and oversampling would be the best. Good $$$ equipment probably does this. Bad filter and no oversample would be the worst. Others are in between.

------------------
Here is a summary of my position. I can only address the BT878 because Nelson would say I was talking about bumble bees if I extended this to an ATI card.

Summary

1. VHS has a limited bandwidth of about 3MHz
2. This equates to a frame about 320 pixels wide
3. The BT878 samples at ~ 754 pixels NTSC and 1135 PAL
4. After sampling, It then does a digital LPF at 6MHz
5. This effecitvely reduces the resolution to 640 pixels max
6. If you ask it to resize the 754/1135 to anything less than 640, you are introducing digital noise due to aliasing. How much there is depends on the source.
7. If you ask the BTwincap driver to resize to anything less than 368, It tells the chip to apply a low quality digital filter at 3MHz
8. This filter probably eats into the 3MHz due to 'roll off' (it is 2 tap. up to 5 tap is available. Got me what that means )

Conclusion:
With the stuff I mentioned above, 368 crop (or maybe post resize) to 352 is probably the best option. 704 is a nice spec number, but you don't get anything xtra but dvd spec. 352 cuts the resolution.


My advice to the orignial poster
Same as I said the 1st time. 320 is the max for VHS. Depends on what your equipment does. Run a test and see. If you don't see any difference, you are wasting your time. In fact, if you can cap and resize/crop, any analog source (short of a test DVD) above about 640 is a waste.

And of course, YMMV.

[Matteo693]

Well, I just re-did a 352x576 capture & whacked it onto DVD - all TMPGEnc settings cranked up to maximum - and it looks like SHITE! There is a BIG noticeable difference from the original VHS source.

If I capture at 352x288 and post-process with Klaus Post's fantastic "Smart Smoother HiQ" filter, then whack it down on VCD, I get HEAPS better results - maybe better than the source??! (= this is a subjective "visual" judgment, not a technical one!!!......

Of course, there is the 80min limit on VCD...I really want to figure out how to capture VHS at full-res & burn to DVD...I realize I'll have to register my PicVideo codec & use that, it seems to work fine, & quality is great...HOWEVER: why can't I seem to get anywhere using huffyuv, or uncompressed avi for that matter??? Surely my system is up to scratch??!

(NB: Yep I've experimented with the huffyuv settings, same result = dropped frames like crazy!)

[trevlac]

If 288 looks better than 576, you probably have a field order interlace problem. Does it look as bad if you play it on the PC? If not, then the field order is probably wrong from your capture.


288 is 1/2 the vertical resolution of VHS.

I don't think many here are going to jump up and say VCD is better than 1/2 DVD.

But let's see .....

[duhmez]

Dedicated captuee cards are not requried ot cap huffy @ 720x480, nor is a second hard drive. Modern hard drives are way more than fast enough to handle full d1 huffy cap, so let's forget this nonsense. ( I use all in wonders., not dedicated capture board, flawless so physically it's possible)

I'd be suspicious it's somehow the card you are using.......
Separate hard drives, dedicated capture cards, Dv... none of this is neede for full D1 huffy.

[Matteo693]

Onya Duhmez! Yes I am inclined to totally agree with you... I figure there's lots of people who are capturing full-res with one HD etc etc...

So it's my capture card perhaps? Yes, perhaps...

Just to remind: I use a :Leadtek Winfast A280 Geforce4 Ti4200 with AGP8X blah blah...

I would be curious to hear from other folk who use this card??? Success stories??? Problems????

One point: When I try to capture VHS with the capture card's software (WinFast PVR) it only allows me a resolution of 320x240. If I change this (to say, 720x576...or whatever ELSE is listed in the drop-down list) then it instantly defaults back to 320x240. I can't seem to change this (on WinFast PVR). Hmmm..... (YES, everything's been updated etc)...

I don't think many here are going to jump up and say VCD is better than 1/2 DVD.

But let's see .....

Well I will. The VCD I produced looks TONS better than 1/2 DVD!!!! So what's going on?

I am even prepared to send a VCD & VHS & DVD copy for anybody to compare... No shit! I wanna nail this bastard once & for all... The VCD is better than the VHS, and the DVD is the worst of 'em all!

(NB: Stupid yuppie copyright legislation etc etc doesn't apply here, it's purely personal VHS footage.)

(NNNNBBBB: But I HAVE tried it on VHS movies as well, JUST to try & crack this bastard!!! == no avail!)

[duhmez]

VCD can look better than 1/2 dvd is the bitrate of the 1/2 dvd res.
For a vhs source, i'd say 2000+ bitrate will look great @352x480(By great I mean very close to the vhs souce.)

one thing you must also do is check your field order, if you do it wrong the video will be ruined.

To check the field order do this:

Selecting the correct field order:
Fire up tmpgenc then do this.
Press the "Settings" button and go to the "Advanced" tab. Under the "Field
order" setting, pick "Top field first (field A)". Highlight the
"DeInterlace" option in the list and double click it to open the DeInterlace
dialog. Select the "Even-Odd field (field)" Method. Left-Click on the video
area and then using the keyboard arrows move to the right to frame-advance
the video. If the motion is smooth, then the field order is correct, if it's
all jumpy then exit the DeInterlace dialog, switch the Field order to
"Bottom field first (field B)", re-enter the DeInterlace dialog and check
that now the movement is smooth. When the correct field order is selected,
make sure that the DeInterlace option is checked OFF! We only use it to
check the field order, it must not be enabled for the final encoding.


If you had wanted to make VCD's from vhs caps, I would always recommend discarding a field instead of deinterlacing. this way the fields don't blend together, you get true progressive frames.