VideoHelp Forum

Our website is made possible by displaying online advertisements to our visitors.
Please consider supporting us by disabling your ad blocker or buy PlayOn (record Netflix) :)
+ Reply to Thread
Results 1 to 3 of 3
  1. Far too goddamn old now EddyH's Avatar
    Join Date
    Jan 2003
    Soul sucking suburbia! But a different part since I last logged on.
    Search Comp PM
    Edit: TLDR version - have a 20ish minute historic documentary, one of three on the same tape, its friends converted just fine from (cinefilm to) VHS to DVD in a consumer tape/disc recorder deck but this one has a field offset and/or display order error causing jerky movement - on what should be progressive material! - and obvious combing even on still scenes. Given my increasingly severe allergy to having to mess around inside the guts of DVD-format video when it's not absolutely necessary, and increasingly obsolete workstation PC, some way of applying software to the ripped-to-HDD VOBs/IFOs would be strongly preferred as I can probably muddle my way through demuxing then doing whatever processing and reconstruction/reauthoring is needed. Have spent better part of a week trying to find an answer independently and have got nowhere, between ancient threads, dead links, and software that *almost but doesn't quite* do the necessaries...


    Greetings fellows! It's been a while. I've not had much time for derping with digital video aside from recording things on Humax PVRs, capturing with good-enough USB DVBT and analogue devices and some set-top DVD recorder stuff of late (well there was the multi-angle Avisynth one-man edit mission fiasco of 2013, but we don't talk about that), and that which I have done was achieved either with simple Joe Consumer grade programs (they're so much better these days!) or Google, so I admit I've been vewy vewy qwiet. But now I am forced to throw myself on the mercy of your knowledge...

    Part of my current job involves the forward transfer of old VHS content onto DVD and/or into MP4 files for whoever rocks up and presents it (usually tutors, as I am at a university), so long as it's at least arguably quasi-legal to do so (ie we can argue fair use format shifting if it's not possible to just get a suitable PAL R2 version from Amazon for a pittance, or stream it using one of the various educational video archive sites). It's a task inherited from a much older guy who pretty much just sat on a bench dozing whilst the land was bought and institution built around him, then woke up to find he was sat in an office with people demanding all kinds of analogue photo/audio/video services from him, and in the couple years immediately pre-retirement busied himself cack-handedly copying huge stacks of old ERA-compliant tapes to low grade DVDRs using LP mode and not quite writing down all of the relevant information in a form anyone else could interpret... ahem.

    Thing is, trying to do it better than that whilst also carrying out other duties is heavily reliant on more automated systems doing what they're supposed to, and that's had a hiccup here.

    Got a tape handed to me by the library, a compilation of British 1940s newsreel-type documentaries focussed around on-the-scene panoramas and soundscapes during the later war years, and a slice-of-life docu-diary thing covering the terminal stages of WW2 and the immediate postwar period. Believe they're probably all public domain now, unless the act of transferring them to VHS about 25 years ago extended the copyright somehow (whoever was responsible for that did a very basic job, simply telecine-ing the original celluloid without the greatest overscan, and providing it with a very plain label and box inlay, so there's not much original IP inherent in that), and I haven't yet found a DVD version.

    It initially looked like that and a few other recent submissions were going to have to go via my established but not particularly efficient VHS to DVD workflow involving a couple different decks, some occasionally iffy cables, rather variable signal levels, a poor choice of quality modes and being careful with pushing multiple buttons on different, increasingly worn-out machines at once, but then a Samsung combined VHS/DVDR deck came available thanks to another department having a clearout. Quite good picture quality, sync-start, and even better a choice of speeds graduated in half hours from 1 thru to 3 hour on single-layer, though the original remote was gone.

    Have transferred a few things with it so far and they've come out reasonably well, apart from sometimes low audio volumes and inability to quite get the tracking entirely clear at both top and bottom (both probably down to the extreme age of some of the tapes, some of them are more than 30 years old and the non-commercial ones were recorded off analogue TV using a 1980s deck of probably very dubious quality) and an annoying ~13 line high area at the bottom of the 576 where the (not entirely brilliant, but serviceable) line-based TBC gives up the ghost and everything goes really squiggly (definitely inherent to the machine as it happens the same with every cassette). But the discs look alright when played back as normal DVDs on both TV and computer, and with suitable cropping (usually to about 680x560) and aspect ratio setting, high-grade deinterlacing to 50p, audio gain boosting and other careful encoding tactics in Handbrake the results come out entirely acceptable for the intended uses. IE we're just preserving investment in study materials, some of which are genuinely unique and irreplaceable, rather than making anything for showing off in a museum or broadcasting.

    The problem has come with this one tape. As there is a precious couple seconds of silent, black dead space between each of the three programmes, and the -R discs I'm currently having to use (as it's a single standard deck, and even if I *had* any -RWs, they're a really awful format compared to either -R or +R/+RW so I'd be loath to use them) doesn't allow any kind of editing other than renaming and deleting, I hovered over the stop button and then made careful with the rewind and pause to put each one on a separate track, which may or may not be the source of the trouble. Yes, I could have just recorded it all in one go and cut it up and reauthored on the computer afterwards, but, seriously, I'm not a student or gainfully unemployed any more. Life's too bloody short, and I've had about all I can take of havinging to be mess around ripping, demuxing, converting, chopping, transcoding, re-syncing, remuxing, menu preparing, authoring, burning.... etc... if there's no need to. ESPECIALLY if it involves DVD format material, which as far as I'm concerned is something that wasn't so much conceived by a panel of so-called industry experts, as mined from a particularly fragrant part of Satan's lower intestine; I've lost track of the man-days lost to making a really nice looking menu set for a carefully arranged set of optimally encoded files only for half of the stuff to either not work at all, or look/sound completely janked once burned and played. If the set-top hardware can do it in a satisfactory manner, and the people who are going to play it think it looks and navigates OK, and it doesn't require any additional hours of thumb twiddling and hoping nothing idiotic happens along the way because of a foible of the format or software, that'll do just fine.

    Anyway, the first two programmes on it are A-OK, there's some speckling and the like which I may enquire about removal techniques for at a later date (as, of all things, VLC's "Yadif X2" deinterlace filter shows signs of being able to remove them - if you play back at 25% speed with it turned on, they appear with the first field of each film frame then vanish once the second comes in, meaning even when played at 100% it has a less corrupted appearance than with the other, 25fps-only filters), but considering that the original films were themselves showing the effects of ~45 years of imperfect storage at the time of transfer, and they'd have shown up if the tape was still in use anyway, it's not a significant problem. It's patina

    The third programme, however - annoyingly the one where the pictures are more important! - has a clear interlacing error. It seems superficially OK on the 14" portable CRT TV I use as a disposable good-enough monitor for the recording and titling process, but when played back on any computer it has not only the classic leapfrog style herky-jerky motion (which is especially weird on footage that, as far as I can tell, was telecined with the film running at a rock solid 25.00fps and so should be effectively progressive), but also obvious combing along diagonal edges, even on stationary images, suggesting that the fields have at the very least been recorded in the wrong "positions", ie top is bottom and vice versa, and are maybe displaying in the wrong order as well (either a separate fault, or a side effect of the first). I somewhat doubt Handbrake's deinterlacer will make a good fist of that... and I don't know if, should I try to crop a line off the top and only trim 15 from the bottom instead of 16, it would apply that *before* any other processing (so possibly fixing the inverted positioning), or *after* (making everything even worse).

    The first thing I want to try here* is directly twiddling the flag bits in the relevant parts of the DVD filestructure and/or the actual MPEG2 video stream itself, if at all possible. As it's hopefully purely a matter of the original video signal having become displaced by 1/50th of a second relative to the DVD encoder vs the other two thanks to the stop-start recording process, this may be enough to fix it...

    *before going down the more well worn but ball-achingly tedious road of pulling the VOBs apart with DGIndex, loading it into Avisynth (and VDub) with DGDecode, using its powerful yet fairly easy to comprehend field-twiddling and motion compensating deinterlacers, then encoding straight to MPG4 from there using VDub (and extracting the stream/inserting a crunched version of the audio with AVIDemux) and re-encoding from the same source to MPG2 with HCEnc, muxing it back together with the audio and either seeing if said VOB can be shunted straight back into the on-disc version of the DVD or authoring a whole new disc with extracted streams from the other two progs and the least unreliable software I can find, then burning with fingers crossed very hard... The inherent water-torture of which is only made worse by the fact of doing it on what is now a very old workplace all-in-one PC with a 2.1ghz Mobile Core 2 Duo processor and fairly limited HDD space (and less of it contiguous) for multiple, "high" speed runs spitting out all the different alternative versions for cross-comparison. This thing is s-l-o-w when it comes to encoding video, and not even that quick when splitting or muxing...

    Problem is, I have so far found NO WAY of actually doing this. There's Restream, but it only works on extracted (and demuxed?) MPEG files rather than VOBs, as far as I can tell. IFOEdit doesn't seem to expose the flag anywhere, at least not in any obvious way. And any other otherwise promising tool or method yet seen have one, other, or both of those two problems, plus some additional stumbling blocks of their own.

    I mean ... come on ... surely there must be a way? It's a simple bit of control data that lets the player know how interlaced content should be output to the screen. One would hope that not only would its location either be known ahead of time, or at least easy to determine from the information in the IFOs and VOBs, but also that, as a one-bit switch (maybe 2~3 bits at most?) rather than some variable length string, it could actually be altered without affecting the rest of the file's structure or playability, or even needing very many sectors to be rewritten. But so far, across the whole internet, with some days of bouncing around both in google and yahoo/bing, scouring forums and program manuals and standards publications etc... nothing. Help!

    (fwiw, there were some potential links over to Doom9, both on the search engines and from certain places here, but their forum appears to either be broken right now (when it was up a few days ago, when I glanced at it and made a mental note to go back and check it out), or silently blocked by our local or's regional firewall. Either way, any address just times out after a couple thousand milliseconds...)

    Apologies for the screed, but if nothing else it'll hopefully nix all the "...have you tried/considered XXXXX?" suggestions which I'd otherwise spend three pages saying "yep, didn't work" or "nope, but I'm not going to unless I really have to" in response to.

    I mean, I probably could have done the demux-avisynth-remux/reencode thing by now, but a) I don't want to admit defeat, b) I'd rather come out of this with a quick and easy method I can use in future should I encounter a similar problem. Either however is better than being That Guy, you know, thhe one that looks after the conversions from DVcam to digibeta for the BBC News and seems to get it glaringly, unwatchably wrong 95% of the time... (and their colleague over at Midlands Weather who simply doesn't understand neither overscan nor aspect ratio...)

    Thanks muchly...

    (PS, side question: even the "OK" videos, of all kinds, seem to have some very minor combing under similar circumstances, kind of like the even lines are horizontally displaced from the odd ones by about one, maybe even one-half a D1 pixel; this I presume is down to an imperfect or at least not fully field/frame aligning TBC in the deck, and wouldn't really be noticeable when played on a TV (especially something with only about 400-500 horizontal pixels like that portable), but I CAN see it on the computer screen, and it's just annoying enough for me to want to see about fixing it. I know it'll require re-encoding, but is there any particular tool that'll sort it with minimal human interference? Or would it be a case of split fields, shift one of them sideways by a certain amount (1.0 or 0.5 pixels, maybe involving a horizontal size double first then shrinking back after), and recombine before encoding again? I might have to see if the same thing persists with running the old standalone VCR to the standalone DVDR, and thus whether it may still be preferable to use that setup for 0 to 60, or 91 to 120 minute recordings, even though the combi would still probably be preferred for 60-90, 120-150 and 150-180 minute...)
    -= She sez there's ants in the carpet, dirty little monsters! =-
    Back after a long time away, mainly because I now need to start making up vidcapped DVDRs for work and I haven't a clue where to start any more!
    Quote Quote  
  2. Post a sample. But it sounds like the top field is stored below the bottom field, and hence the field order is reversed at playback (in addition to the fields being spacially displaced). That's easily fixed with SwapFields() in AviSynth, or the Field Swap filter in VirtualDub. Of course, you'll have to reencode.

    Crop of a normal progressive frame on left, field swapped on right:

    [Attachment 40988 - Click to enlarge]
    Last edited by jagabo; 20th Mar 2017 at 11:18.
    Quote Quote  
  3. I actually did try to read through your entire post. It is extraordinarily long, but contains precious little information.

    I think you are trying to transfer a 1940s documentary; and I think you believe you have some sort of interlacing or combing problem; and I think you can't deal with it because you are capturing using a VHS to DVD recorder and for some reason don't want to use DGIndex and other tools to let you get at the MPEG-2 video and edit it.

    If I have surmised any of this correctly -- and I many not have because you introduced so many tangents -- then the solution is obvious, although it may not be what you want to hear: you need to capture the video to an editable format because, as you have noted, the VOB structure on a DVD is not a very friendly format to edit. Then, you need to post 10-15 seconds of this video (preferably a section containing lots of motion), without re-encoding, so that we can actually see what you are dealing with. Reading between the lines in your post, I assume you live in Europe and I assume this is PAL content. The documentary was obviously shot on film (no video in 1940). As you well know, that is usually just transferred from 24p directly to 25 fps interlaced by speeding it up slightly. In the old days, they wouldn't bother to correct the pitch. Thus, the usual way to restore this is to change the fps header in the captured file from 25 to 24, the pitch will take care of itself, and your work is done. Most modern equipment will play native 24 fps just fine, whether that equipment is NTSC or PAL.

    However, there are a thousand ways this video could have gotten screw up, and this includes format conversion to NTSC, and then back again. There are so many different things that can happen that, without a sample, it will be impossible to offer specific advice. The usual AVISynth tools are TIVTC (TFM/TDecimate), SRestore, along with custom scripts, depending on what other stuff is going on.
    Last edited by johnmeyer; 20th Mar 2017 at 11:24. Reason: added "lots of motion"
    Quote Quote  

Similar Threads