I'm a filmmaker who has a pile of projects on NTSC VHS's that I want to preserve into big (interlace-kept) avi's in high quality. I've been reading on this and other forums for years about doing such things; have bought the equipment and been doing learning tests off and on for awhile now. Gosh, there are so many things to learn about and keep track of with this! Anyway, my first question that I haven't found an answer to in my knowledge-aquiring/trial-and-erroring/googling...
When I do multiple captures of the same scene, the field order seems to be random initiated (<- my guess anyway). To explain: in viewing the resulting supposed almost identical captures in VirtualDub, half of my captures of the scene will have cut x be a clean "progressive" style cut from one shot to the other while the other half of the captures of the same scene will have that same cut interlaced. (I also notice that one style is always a scanline lower than the other.) I basically understand what is happening I think. At each starting of the capturing, the capture card is kicking in on either the top field or the bottom at its start, resulting in two different looking types of captures in the end, right? But I ask... a) Is one of the two results more "correct" than the other? (I would assume that the one shifted down a scanline might be the "wrong" one if there is such a thing) b) Is there a way to have all the captures turn out the same in this regard, a setting in VDub or something? (I am wanting to do to multiple captures and carry out the "averaging" technique with Avisynth, but as it is I have to throw out half my captures because averaging both of these together will create a mess as I want to keep things interlaced).
Sorry if these turn out to be rather basic questions, but I promise I've been trying to seek the answers on my own and they've just eluded me so far. Thanks for listening.
Rough capturing approach specs: Hardware - AMD Athlon 2100+, PDI Deluxe card, JVC 9911U S-VHS Software - VirtualDub or IuVCR for capturing, HuffYUV codec, Avisynth 2.5 (for averaging after capturing)
For my old analoge interlaced home camera footage I used a Canopus device that converts the analoge material to DV and the results were (as far as I know) always the same. But I'm from PAL country and don't know much about NTSC. But it sounds like whatever field enters the capture device first is turned into a top field. So if that field is already a top field nothing changes but if it is a bottom field that would explain the shifted scanline (vice versa if device starts with bottom field). Is that what you meant by "kicking in at either top or bottom field". This sounds like a question for the doom9.org forum unless someone will come up with an answer .
Edit: this question is getting more interesting the more I think about it. My captured DVs are always bff (I think all DV is), so that means when capturing starts the device starts at the first bf it encounters or field order is changed if it starts at a tf. Fez: I am so excited about Star Whores. Hyde: Fezzy, man, it's Star Wars.
Is that what you meant by "kicking in at either top or bottom field".
Yep, that's what I meant. I just assumed when I started all this that the capture card would decide at its starting to always do the (field) capturing one way, the "best" way -- or that at least I could click a setting somewhere to force it to start the capture only when the requested field passes through its clutches, the way your canopus appears to do (according to your edit text above). As you may know, the PDI Deluxe capture card is well regarded, 'twas used for the X0 project, so I can't imagine this 'problem' I'm having as being a real one. Did the X0 people really have to do multiple captures to get ones that "took" right and disgard the ones that were the wrong field first? Or is there another piece of this puzzle I don't know about? I'm just not finding that friggin' control/option select which I think must be there somewhere!
I know they could probably answer this at doom9.org (where I've read gobs of stuff over the years of course), but I feel I have already have too many memberships on bbs's going, and my signing on here has to do with more than my wanting to ask one technical question about video capturing -- but I'll keep that reason a secret for the moment I think. If after a while no one has an answer I'll go over there, but I'm not in a huge hurry on this project or anything.
Is there a way to turn off field capturing and switch to frame capturing?
Not than I can see. In VirtualDub there is "Swap fields" but in looking that up and trying it, that is not of any help (it really screws the captures up, a fix for something that's not broken with my particular capping set-up). I did more searching around the internet just now and still found no one else having talked about this particular bizarreness. At least my captures all look great and beautiful, and for that I am thankful. I could just discard ~half my caps, do only an averaging of the "shifted down one line" versions I guess, but the learning side of me would really like to know what is behind this weirdness rather than give up and do simply what works in a state of ignorance. (<-not to mention the time of having to do up to 10 or 11 captures instead of a straight 4 or 5 for everything I wanted to do averaging on!)
I did more tests to make sure I wasn't imagining things, and the same still holds. To risk being repititious (just hoping I'm doing helpful clarifying), doing a slew of caps of one scene will result in two "categories", the interlacing starting point (and flowing 3-2 style) at a different point between the two but consistent within each, and one of the two varients is lowered one scanline compared to the other (i.e., all of the captures that are in the "shifted down one frame" category will have an identical interlace pattern throughout the footage and are completely "average-able" barring an auto-inserted duplicate frame every few minutes).
Thanks again for the replies and trying to help me look into this Arnie.d.
When you say "3-2 style", are you referring to 3:2 pulldown? If so, I assume you will be doing IVTC before averaging. Therefore, it doesn't matter which field the capture starts with - the IVTC process will re-arrange them to give a progressive 23.976fps output. So there is no need to discard half your captures!
Guidelines for post content and general behaviour: read announcement here
Max. allowable image sizes in signatures: reminder here
Actually, I've been doing my averaging without IVTC'ing so far -- yes, interlaced averaging -- and it works/looks fine (within one of the two aforementioned groupings).
I know that IVTC'ing will wipe away the 3:2, but I'd still need to shift one group up or down a line to make them match up. While I know that's not hard, I'd still like to see if I can find a way to just not have this difference happen with my captures in the first place (if that's even possible). Do you happen to know if what I'm experiencing is standard or if it's strange? Maybe it's unique to my set-up due to my having a bad setting somewhere and I can fix it. That's my hope I guess.
While I'm here.... Should I capture at 29.970 or 29.976 fps? I have yet to find a web page with the cases behind those spelled out (feel free to just give me a link).
I thought IVTC is ONLY to remove pulldown, not to turn something that's interlaced into something progressive (unless the source was progressive). Now I don't know exactly how it is for NTSC camcorders but for PAL camcorders the camcorder while filming doesn't capture a frame and turn it into seperate fields. It's capturing fields at 50 fields per seconds. Because of this it is hard to turn interlaced camcorder footage into progressive footage (because the two fields are not shot at the same time they can differ a lot especially in faster moving scenes). But as I said I don't know how it is for NTSC camcorders. Fez: I am so excited about Star Whores. Hyde: Fezzy, man, it's Star Wars.
Moth3r: I see in my Device Manager that I have two entries for my PDI Deluxe card (both with exactly the same wording). Digging into the "Details" of each of those two, going through the list ("Device Instance ID" "Matching Device ID", etc), I find that a what they say in the big white field does not always match from one to the other. (One's sting of stuff will end in "48" and that other in "49", things like that.) But I notice that in "General/Location" or their two properties pages, one entry reads "PCI bus 0, device 9, function 0" and the other reads "PCI bus 0, device 9, function 1". Is this just a complicated device that needs two Device Manager entries, or does this point to me having screwed up on installing the driver(s) somehow? This is particular area (deciphering device driver details) is not my area of expertise at all I freely admit. Is there a better place to look for that setting ("frames or fields") than here in the device manager?
And 29.970. Cool to finally have that setting nailed!
Arnie: (Putting very briefly) With NTSC, only some of the frames are interlaced, the others look progressive. So IVTC for NTSC does both some de-interlacing and frame dropping (plus the frame rate change) in the correct patterns.
The footage I will be transferring down the road (after I have adequate knowledge and experience) is indeed from sources that I will want to keep interlaced: Camcorder stuff, super-8 films footage transferred in ways that did not assure 24fps accuracy, would not be good candidates for doing ITVC. Hence, I wish to just leave most if not all of it interlaced. For my tests right now, I'm obviously not using (wearing out) those tapes; just more disposable stuff.
Yeah, I guess NTSC camcorders record in tru 29.970 interlaced and it cannot be IVTCd since the "source" was never 23.976 fps progressive, correct? Fez: I am so excited about Star Whores. Hyde: Fezzy, man, it's Star Wars.
I haven't captured any of my VHS camcorder footage yet, but as far as my understanding goes you hit the nail right on the head. It's shot right into the result that telecine'ing works to bring about in film-transfer sources.
Is there any way you could post screencaps to better illustrate what is happening? I have the pdi card and I have had problems too, the video image is shifted up in the capture. My audio professor said it might be a timing issue. http://forum.videohelp.com/topic335815.html
But that's not the same thing you're seeing, right? I've never noticed anything wrong with the fields in Sony Vegas. Even if I switch between even & odd fields first in the preferences or render box. But then again I haven't worked much with the pdi card so far. Either way, I have got to believe that there is some kind of control or plugin to compensate at the capture stage so you don't have to average it out in avisynth.
Thanks for the reply KM. Cool that someone else with a PDI Dlx card is here!
I actually saw your threads weeks ago in my googling around for solutions, your talking about that hugging-the-top/uncentered issue . No, it's not the same thing -- and I, too,wouldn't mind having the captures moved down a couple of lines at least. Too bad no one responded to your thread.
Posting screen captures here wouldn't show nearly as much as your seeing two short avi clips compared overlapping style in VDub... but still, here's pics of exactly the same frame of video capture...
I captured the clip 8 times. Most of the the caps turned out with this cut from the one shot to other looking like this: [url=http://www.freeimagehosting.net/][/url]
Okay, that cut is interlaced. Fine. However, a rebel group of the captures have the same cut clean from the one shot to the other (this here being the "cut to" frame of course). And all of the captures in this style are one scanline lower than in the capture style above: [url=http://www.freeimagehosting.net/][/url]
Theoretically, all the captures should look basically identical except for very slight noise changes, but that's not what I get. (I won't describe it again as I've already gone through it in detail twice above in posts.)
...to compensate at the capture stage so you don't have to average it out in avisynth.
Note that my wanting to do averaging has nothing do to with this problem. I plan on doing averaging with all my captures, especially after I get this sorted out. I believe averaging is a way to make a capture better (by combining a number of them together to fade out much of the noise before using any de-noise filters!). If you have averaged all the noise down to a fraction of a percent of its original level in the capture area, then you don't have to push any denoising filters nearly as hard, which is a good thing for keeping the quality intact of course.
Knightmessenger, have you done multiple caps of one thing and gotten the differences talked about here in those captures?
No, I haven't come across anything like that. Is that on a frame right before/after a cut? What vcr are you using, by the way? And how good of shape are the tapes in? Do you think that they could just be a little worn out that the tape does not play precisely the same every time and that the capture card is merely faithfully capturing that? I know the format is not the most frame accurate or warp proof.
Thanks for the reply. It's the frame after the cut. I could have put the frame before the cut instead I suppose but same difference. Basically, one capture style interlaces that cut, the other keeps it clean for the two shots (and lowers everything a scanline).
Regarding VCR: It is a JVC HR-S9911U -- one of the top consumer models made -- so it's no slouch in the VCR department, and mine is in very good condition. No indications of anything wonky with it.
Regarding tapes: Every single tape of the dozens I've tried results in this. Some of them are barely played pre-records, so it's not a tape problem I'm sure.
The cable (in case that's what you might suggest next) is a $150 S-VHS Monster cable, going straight from the VCR to the PDI card. So I doubt it's that either (especially as I've used more standard cables also and no difference).
In a narrowing down fashion, this leads me to thinkin' that perhaps it's my PDI card drivers? Do you still have a copy of the sweetspot drivers you used to do a compare? This is what I used/installed: sweetspot_wdm_driver.exe, the file size (in properties) listed as 524,288 bytes.
Edit: While I had confidence it was not my JVC9911U that's the prob, for kicks I hooked up my "cheap" VCR and did a bunch of captures with that. Interesting find... I did 9 caps in a row of the same segment above, did each by hitting f6 then after three or four seconds hit ESC, back-scanned, then hit f6...ESC, scanned back the few seconds, repeated this 9 times to get my nine capture files. All nine came out exactly the same way (like the second pic above). "Culprit found!" I thought, "'Tis my beloved 9911U that's the problem!" However, I did some more caps for kicks, this time hitting stop on the VCR and rewinding after each (rather than just back-scanning) and I got the same problem I've been having, half the caps like the first pic above and half like the second. Note that I never turned the VCR off, nor did I exit the "capture" screen in VDub for these. Whatever this change is, hitting the "Stop" button (or "Play"?) on the VCR is all it takes to allow it to happen. Curious.
I don't know exactly what you mean by sweetspot drivers. I probably have them somewhere but I didn't actually install the card so I wouldn't know where to find them and what to look for. I'm totally annoyed about the image shift issue though. I mean, how could no one know on video help? (almost 100 views, no replies) It's good that you discovered a way around it but that does seem weird that pressing stop would make a difference. Maybe the heads somehow can't hit the same field each time. However, I actually did encounter that example of a interlaced cut but it wasn't quite exactly with video. And I'm pretty sure it will give me the same result at that frame every capture too. Just for fun here it is. [URL=http://imageshack.us][/URL] I must not have paid attention to what it actually looked at full size so I didn't realize I would get a double image. Would you have any suggestions about where else to ask about the PDI image shift?
Would you have any suggestions about where else to ask about the PDI image shift?
I saw a thread that went into great detail about inaccurate aspect ratios. Did you know that different capture cards can render the captured the video into the frame at different widths even though it's all supposed to be the same? I.e, Cap card A may make the picture area of the VHS capture 704 pixels wide in the final avi, but Cap Card B might make the same picture area only 699 pixels wide (but all will still give a 720 wide image stream in the end of course). I found that a little shocking personally, but such "play" being possible makes sense when you think about it. Anyway, the reason I bring it up now is because they talk about how to fix your capture's width or height if either is too narrow or too wide. I'm thinking that maybe you (and I) can use this same adjusting tool to simply lower the captured image area within the frame.
After digging, I found the thread... http://www.doom9.org/capture/capture_window.html
Toward the bottom they talk about making adjustments for bt878 cards which, luckily for us, the PDI Deluxe is a variety of. Again, I haven't done this procedure yet, but it's on my to-do list. A pretty cool idea, wish I'd thought of it: burn a DVD with this test pattern then capture your DVD's player's output of this burned disk and see if it's wider or narrower or what. Neat!
Well that's kinda what I was looking for but I'm more interesting in moving the image down, not shrinking or stretching it. It's also a bit more advanced than what I can understand. Thanks for the help though.
Yeah, I know you're only interested in shifting the image down, but I was thinking just one of those little settings might do just that for you. Relooking at that thread, its complication seems to come from all the calculating to figure out stretching the image. If you posted on Video Help something like "BTTool and PDI Dlx card -- I just want to lower the image cap area in the frame!" someone might be able to tell you the easy/specific way to use the tool for only that. Just a thought. I would mess around with it now myself, but I have a lot of other things to juggle atm.
"I would mess around with it now myself, but I have a lot of other things to juggle atm."
Same here. Besides, wouldn't that lose resolution if I shrank the video.
EDIT: Then again, reading what you wrote. "Cap card A may make the picture area of the VHS capture 704 pixels wide in the final avi, but Cap Card B might make the same picture area only 699 pixels wide (but all will still give a 720 wide image stream in the end of course)" maybe not? Somehow I get the idea that a 699 pixel capture could never gain back the 21 pixels.
maybe not? Somehow I get the idea that a 699 pixel capture could never gain back the 21 pixels.
I'm not sure we're quite on the same page, but I think it's my poor wording. I'll try again, a different approach... The resulting avi from a capture that is set to 720x480 will always be 720 pixels wide. Now, what if the capture card and/or its driver is so screwed-up up that it's taking the VHS picture and making it only 100 pixels wide for your "video capture area" width within the full capture frame? Each frame is still 720 wide in the end, but within that 720 width you have 620 pixels of pure black and 100 pixels of really squeezed video from your VHS.
A capture card is capable of putting the wrong width/height image from the VHS into your 720x480 avi, and BTTool seems to be a way to fix that, that's what I'm trying to say. Obviously, a picture area that is only 100 pixels wide (within the full 720 pixels image) is erroneous, but what if it's 6 pixels off from what it's supposed to be rather than hundreds of pixels off? You might not notice a 6 pixels squeeze with your eye, but it's still wrong and it's good to correct it at the capture stage if possible. That's what the doom9 page I posted a link to was taking about. I'm just piecemeal regurgitating from that page, and I bow to its expertise so feel free to go back there for more straight info.
I believe the ideal capture area picture width from VHS is 702, leaving 18 pixels of black to the left and/or right of that 702 to fill out your 720 width. You and I are more worried about the top of our captures because it looks like there might be cropping happening at the top of the frame. At least that's what I'm worried about, and want to correct. Still have too much stuff going on right now though
I think that's what I mean. I wonder if the same thing happened to the X0 team with their captures. But luckily for them, they would only lose part of the black bars.
If after a while no one has an answer I'll go over there, but I'm not in a huge hurry on this project or anything.
At least my captures all look great and beautiful, and for that I am thankful. 
[/url]
[/url]
[URL=http://imageshack.us][/URL]