VideoHelp Forum
+ Reply to Thread
Page 2 of 8
FirstFirst 1 2 3 4 ... LastLast
Results 31 to 60 of 212
Thread
  1. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    No complaints about soft image in the IVTC clip. Must've been an optical illusion, because looking at it later showed it to be more solid than the bob'd or yadif versions.

    But there are still 7 highly visible Christmas balls and raspberries in the IVTC clip. The whole point of the exercise was getting rid of those pops and blisters. For color correction, repairing frames, etc., the VHS material is cut into more than 150 small AVI's, rendered separately to MPEG, some using TMPGenc, some using CCE, then patched in and joined in TMPGenc Editor. Or I could take the existing MPEG compilation-so-far and IVTC that. Still seems like an awful lot of trouble to IVTC and re-encode 1.5 hours of 150+ video segments just to get some noise out of 1 minute of titles.

    Just had another idea I can try later tonight. That yellow light bulb is still sitting on Jackie's chin.
    Quote Quote  
  2. Originally Posted by sanlyn View Post
    24fps progressive will play on all NTSC DVD players and CRT TV's?
    If you add 3:2 pulldown flags. Most commercial DVDs are made this way. The pulldown flags tell the DVD player how to create 59.94 fields per second from the 23.976 fps (or other rate) progressive frames.
    Quote Quote  
  3. Member
    Join Date: Sep 2007
    Location: Canada
    Search Comp PM
    Originally Posted by sanlyn View Post
    That yellow light bulb is still sitting on Jackie's chin.
    Use removedirtmc, it's more selective / less damaging than vanilla removedirt and it will pick up the yellow chin spot on an ivtc'ed base clip

    If you're are really picky, what people do is composite different versions and mask out sections or part of frames. e.g. You might use a base lightly filtered version that retains details like lion hair etc... but it might miss a few spots. So you circle spots on the strong filtered layer, so it covers up only the spots. Very easy to do in something like after effects

    A weak version might use 20 , a stronger version that picks up more spots might use something like 60-80. But be careful, important details start to get eroded at higher settings. Usually you want to use as weak possible settings to preserve as much quality as possible

    e.g
    removedirtmc(20, false)


    Code:
    function RemoveDirt(clip input, int limit, bool _grey)
    {
      clensed=input.Clense(grey=_grey, cache=4)
      alt=input.RemoveGrain(2)
      return RestoreMotionBlocks(clensed,input,alternative=alt,pthreshold=6,cthreshold=8, gmthreshold=40,dist=3,              dmode=2,debug=false,noise=limit,noisy=4, grey=_grey)
      # Alternative settings
      # return RestoreMotionBlocks(clensed,input,alternative=alt,pthreshold=4,cthreshold=6, gmthreshold=40,dist=1,dmode=2,debug=false,noise=limit,noisy=12,grey=_grey,show=true)
      # return RestoreMotionBlocks(clensed,input,alternative=alt,pthreshold=6,cthreshold=8, gmthreshold=40,dist=3,tolerance= 12,dmode=2,debug=false,noise=limit,noisy=12,grey=_grey,show=false)
    }
    function RemoveDirtMC(clip,int limit, bool "_grey")
    {
      _grey=default(_grey, false)
      limit = default(limit,6)
      i=MSuper(clip,pel=2)
      bvec = MAnalyse(i,isb=false, blksize=8, delta=1, truemotion=true)
      fvec = MAnalyse(i,isb=true, blksize=8, delta=1, truemotion=true)
      backw = MFlow(clip,i,bvec)
      forw  = MFlow(clip,i,fvec)
      clp=interleave(backw,clip,forw)
      clp=clp.RemoveDirt(limit,_grey)
      clp=clp.SelectEvery(3,1)
      return clp
    }


    If you wanted to fix the sides of the frame, one approach is the mirror the edges so it extends into pure black pillarbox. But a pure mirror will get some of the film ribbon on the right side, because the distance isn't big enough . You can use CC Repetile in AE for this, but I'm sure there are ways in avisynth as well . But I prefer what you did to it instead of trying to match it up.

    On a IVTC only of the Lion section , there are 4 frames that have residual combing. You can use vinverse() on those 4 frames. But vinverse is pretty selective (not too damaging), so many people just use it on everything for ease of use instead of applying it to selective sections

    Or I could take the existing MPEG compilation-so-far and IVTC that. Still seems like an awful lot of trouble to IVTC and re-encode 1.5 hours of 150+ video segments just to get some noise out of 1 minute of titles.
    Exactly. If you've already done all the work, it doesn't make any sense to go through it again
    Last edited by poisondeathray; 12th Nov 2011 at 16:43.
    Quote Quote  
  4. Member
    Join Date: Sep 2011
    Location: NTSC
    Search Comp PM
    It's better to encode as progressive with pulldown, better for bitrate, and better playback from upconverting players. Just wanna clear up some confusion here, your VHS is progressive 23.976 content encoded as 29.97fps by repeating the image in a regular pattern. You can throw away the repeated image and make the rest progressive.

    Fields
    Video is stored as the even lines at one instant of time (called top fields*), and odd lines at a different instant of time (bottom fields). If you make top and bottom come from part of the same image at the same instant in time, you can decode this just by waiting two time periods and combining the top and bottom fields together into a progressive image. Truly interlaced material is not from the same instant of time, the problem here is each instant of time is from a different place in space. You have to guess the in-between lines for each field to make them progressive at ~60fps. (59.xxx).

    Review of pulldown process
    Film at 24fps has to be stetched by 5/4 to 30fps video. We do this by repeating some fields. This diagram shows the repetition. It is drawn in units of 1/120s; in this case a film frame lasts 5 units - 5/120=1/24, and a video field lasts 2 units, 2/120=1/60. This is shown as "0...." meaning film frame number 0 lasting for 5 units. The other lines are made by taking a snapshot of the film frame at that instant for the two fields of NTSC video.

    Code:
    film 0....1....2....3....4....5..
    top  0...0...1...2...3...4...4
    bot  ..0...1...2...2...3...4...5
    fram cln lac lac cln cln|cln lac
    cln means clean, or you see a progressive video frame at this point, and lac means interlaced, you see combing on that video frame number.

    So in other words, starting from the beginning, and stepping through the video one frame at a time, you will see:
    video(0)=film(0), video(1)=film(0,top)+film(1,bot), video(2)=film(1,top)+film(2,bot), video(3)=film(2), video(4)=film(3), video(5)=film(4)...

    What is IVTC?
    So we've got film(0), film(2), film(3) and film(4) ok, we just need to reconstruct film(1).
    We do this by combining parts of video frames. film(1)=video(2,top)+video(1,bot).

    How is that different from deinterlacing?
    From the description of fields, we know that deinterlacing is filling in the missing lines of each field. In the case of deinterlacing pulldown video, it will be a total mess. In video(0) we take the top field and guess the odd lines, however we already know the odd lines, but instead we throw the real odd lines away and guess at them. That's why the result will look blocky.

    Combined IVTC and Deinterlacing
    There's some reasons that our pulldown pattern is screwed up; for example your capture card missing a frame, or the video was edited as video destroying the pattern (for example, an edit point was on an interlaced frame, which means we no longer have all the fields to reconstruct).
    In this case, automatic algorithms recognize the problem and do a deinterlace for this video frame. The result, a brief flash of blocky video at worst.

    The name 3:2 pulldown: see there's 3 0's above followed by 2 1's, then 3 2's and 2 3's, so it's a 3:2 pattern, 3 fields of one film frame and 2 fields of a film frame.

    *I'm not getting into the technicalities of the terms here; really I should be saying field 1/2 and top frame has more to do with which field comes first in an encoded file.
    Last edited by jmac698; 12th Nov 2011 at 16:59.
    Quote Quote  
  5. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Thanks, jmac698, I found that table informative and have notes on websites that go further into it. (Saints alive, do we have time for this yet?)

    I can avoid interlace altogether by running RemoveSpots and/or RemoveDirtMC using SeparateFields, so those overlay routines of mine will work on any of these configurations (Separate Fields is needed only for interlaced) to get rid of the 7 surviving booboo's. I'll look into ITVC for the whole thing. As jmac698 noticed, and as I just noticed going thru later scenes in the VHS, some of the VHS is 23.976 upped to 29.97, and some is just straight 29.97 that always looked by its contrast and color to be from a different source than the rest of the video.

    Will get up an IVTC'd and re-rendered DVD of some scenes done so far from VHS and see how it plays. Might be tomorrow before I can burn it to disc.
    Quote Quote  
  6. Member
    Join Date: Sep 2007
    Location: Canada
    Search Comp PM
    Originally Posted by sanlyn View Post
    I can avoid interlace altogether by running RemoveSpots and/or RemoveDirtMC using SeparateFields, so those overlay routines of mine will work on any of these configurations (Separate Fields is needed only for interlaced) to get rid of the 7 surviving booboo's.
    Sorry to keep pestering, but the yellow chin spot won't be removed if you do that . The reason is the duplicates. It's not detected as a "spot to remove" because it appears the same in 2 adjacent fields . This is hard telecine (encoded as interlaced), which means the duplicates are actually encoded and take up actual bitrate as opposed to soft telecine (which has no duplicates and it's encoded as progressive, but repeat field flags are used to output 60i signal)

    You really have to IVTC if you want progressive temporal filters to work properly . This is a perfect example that demonstrates this nicely. You can avoid "interlace" (as in method of encoding) if you IVTC. It's not interlaced to begin with, it's only encoded interlaced, but the content is progressive.

    Separate fields is usually used for interlaced content, which is encoded as interlaced ("content" and "method of encoding" are 2 different things). If this were the case, you would have to group even and odd fields before filtering - if you recall that was necessary in the opera example because of the field offset. Filters that have an "interlaced" switch automatically do this
    Quote Quote  
  7. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Nope. We're back to those 150+ cut and edited segments again. Let's say I have segment A, it's 832 frames. The last 20 frames aren't used, they exist only to prevent audio dropout when two segments are cut and joined, so they're discarded. I've saved every TMPGenc-Plus .tpr workfile and every TMPGenc Editor file that created, cut, and joined those segments. I can IVTC that segment A, but it won't have 832 frames any more. The editor will make a mess of it, because it formerly cut that segment at frame 812 - which won't exist after conversion to progressive. Some of it was with CCE, which I no longer have installed (I really dislike that app).

    OK, time to think about this.
    Quote Quote  
  8. Member
    Join Date: Sep 2011
    Location: NTSC
    Search Comp PM
    According to my diagram, using separate fields you'll get 3 frames that look the same following by 2 different frames that look the same. The problem here is the denoisers expect each frame to be different. So that's not a simple solution. IVTC really should be the first step.
    Quote Quote  
  9. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    I'm referring to extracting a patch of a few "clean" pixels to overlay onto the same area in another field.

    Even the progressive version still has 7 obvious bloopers in a short sequence. For that version, I would need neither SeparateFields nor bobbing. I'd overlay a clean patch over the offenders. The 4x4 or 8x8 pixel patch would be identical in appearance to the same spot on either a preceding or following frame, and becomes too small and quick to notice during play.
    Quote Quote  
  10. Originally Posted by sanlyn View Post
    I can avoid interlace altogether by running RemoveSpots and/or RemoveDirtMC using SeparateFields
    SeparateFields alone should't work. The spots appear in both fields. SeparateFields will turn the two fields into consecutive frames. RemoveDirt/Spots will see the spot in two consecutive frames and assume it's part of the picture. The following:

    SeprateFields()
    evn=SelectEven().RemoveSpots()
    odd=SelectOdd().RemoveSpots()
    Interleave(evn,odd)
    Weave()
    shouldn't work well either because some of the spots will be in the repeated fields.

    Originally Posted by sanlyn View Post
    some of the VHS is 23.976 upped to 29.97, and some is just straight 29.97
    That's not possible for a film source. I suspect your 29.97 sections have a repeated frame every 4th frame. The duplicates can be removed to restore the clips to 23.976 fps.
    Quote Quote  
  11. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Originally Posted by jagabo View Post
    Originally Posted by sanlyn View Post
    I can avoid interlace altogether by running RemoveSpots and/or RemoveDirtMC using SeparateFields
    SeparateFields alone should't work. The spots appear in both fields. SeparateFields will turn the two fields into consecutive frames. RemoveDirt/Spots will see the spot in two consecutive frames and assume it's part of the picture.
    I know what you're saying. I might not have described my intention correctly. I was referring to the overlay routine I used, borrowing a clean patch from another frame and placing it over the noise. Rather than bob to do this, I can separate fields (on interlaced video) and overlay the fixed patch onto the offending area in the same way I did with the bobbed frames. Of course I'd have to use the same overlay routine on both even and odd fields.

    Originally Posted by jagabo View Post
    Originally Posted by sanlyn View Post
    some of the VHS is 23.976 upped to 29.97, and some is just straight 29.97
    That's not possible for a film source. I suspect your 29.97 sections have a repeated frame every 4th frame. The duplicates can be removed to restore the clips to 23.976 fps.
    Oops. I didn't describe that very well, either. I mean to say that some sections show a different pattern. There are duplicate images, just not as many and not in the same pattern: not 3 dupes of every other frame, but 2 frames in every 5.

    Masking: I have a couple online tutorials for AE showing how a color layer is loaded behind the main image, then a mask is made of the main image with subtractive cutouts that allow the underlying color to show through. So I'd "cut out" everything in the lion frames except the lion, the ribbon, the emblem, etc., and let RGB16 black show thru. But I'm still not up with AE yet to actually do it.
    Quote Quote  
  12. Originally Posted by sanlyn View Post
    Originally Posted by jagabo View Post
    Originally Posted by sanlyn View Post
    some of the VHS is 23.976 upped to 29.97, and some is just straight 29.97
    That's not possible for a film source. I suspect your 29.97 sections have a repeated frame every 4th frame. The duplicates can be removed to restore the clips to 23.976 fps.
    Oops. I didn't describe that very well, either. I mean to say that some sections show a different pattern. There are duplicate images, just not as many and not in the same pattern: not 3 dupes of every other frame, but 2 frames in every 5.
    I'm not exactly sure what you're saying, but sounds like it might be a progressive PAL source converted to interlaced NTSC with hard 3:2:3:2:2 pulldown. That can be restored to 25fps progressive and 3:2:3:2:2 pulldown flags added with DgPulldown without changing the running time.
    Quote Quote  
  13. Member
    Join Date: Sep 2011
    Location: NTSC
    Search Comp PM
    I agree, sounds like PAL->NTSC. srestore can fix that to progressive. We need a several seconds sample to check.
    I don't know what to do about the joining, you can join in avisynth too, if you just list the files in order.
    Quote Quote  
  14. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Originally Posted by jmac698 View Post
    I agree, sounds like PAL->NTSC. srestore can fix that to progressive. We need a several seconds sample to check.
    I don't know what to do about the joining, you can join in avisynth too, if you just list the files in order.
    Not that easy. The cuts were more complicated, and there's no one-step/single bitrate/one-color-setting solution for this critter. Like every VHS movie, color grading between scenes was a nightmare, much of it done in TMPGenc.

    IMHO its original source was likely a soiled theatrical release print (or, put it this way: how does PAL digital video get hair on it?). Likely digitized in the UK, since Americans have little patience for films that aren't action, toon, anime, or porn. How do we know it didn't go directly from film to digital to NTSC? I mean, what sort of signature would differentiate this between PAL->NTSC and FILM->NTSC? But there was definitely a PAL version, as this film was very popular in Europe, especially in France and Germany.

    I'm gearing up to work 2 versions: finish the one that's nearly done, in standard NTSC 29.97 fps, then start a progressed 24fps/pulldown version -- that is, if my 24fps test this week runs properly on non-progressive-scan players and CRT's. I'm not the only one who'll be watching this video.

    Will try to upload a bit of the original VHS capture.
    Quote Quote  
  15. Originally Posted by sanlyn View Post
    I mean, what sort of signature would differentiate this between PAL->NTSC and FILM->NTSC?
    jagabo already mentioned the difference. The former could be 3:2:3:2:2 and the latter 3:2. I find the chances of the former remote, as I have yet to see a retail DVD sourced from PAL and converted to NTSC like that, but maybe VHS tapes from PAL sources did that kind of thing. But, for the third time, a sample will tell us what you have and how it should be dealt with.
    But there was definitely a PAL version, as this film was very popular in Europe, especially in France and Germany.
    Movies shown in movie theaters in PAL countries are shown at 24fps, just as they are everywhere in the world. Or maybe you talking about home video versions available in PAL countries? I can find no record of a DVD of the film ever being released anywhere. Maybe only VHS tapes are around.
    I'm gearing up to work 2 versions: finish the one that's nearly done, in standard NTSC 29.97 fps...
    A mistake, in my opinion. 29.97fps is standard for NTSC DVD, but not for film. That is, 29.97fps is not the 'base' framerate of film.
    ...that is, if my 24fps test this week runs properly on non-progressive-scan players and CRT's.
    jagabo and I both said it would. It won't even author if it's not DVD compliant. And once authored, any NTSC DVD player ever made is supposed to be able to play it. DVD is basically an interlaced medium anyway.
    Quote Quote  
  16. Member
    Join Date: Sep 2011
    Location: NTSC
    Search Comp PM
    Could be another misunderstanding here. If you're knowledgable you make all sorts of assumptions and don't think to explain them.

    There's two kinds of interlaced DVD's
    One is called soft interlaced. This is 24fps with markers on each field to say, display this field again. It's called a repeat field flag.
    The other way is hard interlaced. This is 30fps* video without any flags. Of course, nothing is preventing it from having repeated fields. The player just plays it, and doesn't know it's repeating the same field.
    The third kind, is hard interlaced without any encoded repeating fields. This is true interlaced video. It's stored the same way as the 2nd kind, so they are more sub-types.

    So what's the difference?
    A DVD player faced with a progressive 24fps video, with repeat field flags, will repeat fields of the video on playback*2. If it's hard interlaced, it won't repeat any fields on playback, but play the fields as they are (which could *happen* to have encoded, repeated fields). The output is no different, and a modern TV can recognize if there's repeated fields regardless and create progressive content (however it is a guessing game to recognize the pattern, sometimes it can screw up).

    An upscaling DVD player can simply skip the step with the repeat field flags and display the raw progressive output. If it's hard telecined, the player has to guess at the pulldown pattern. The difference here is that guessing can mess up sometimes. It's better that the DVD is playing the raw progressive video.

    The Encoding Makes a difference
    You can encode 24fps with flags (literally a few bits each) to indicate which fields to repeat, or physically repeat the video and encode that. Although repeated video can encode fairly well, it's still wasted bits. The essential difference is encoding phsyically repeated video, or simply indicating fields of video to repeat on playback.

    IVTC During processing
    IVTC should be the first step in processing, because many algorithms assume that each frame is different and work based on that, for example removing a spot that occurs in only one frame. If you feed it repeated fields, it could see the spot on two fields in a row and not recognize it's dirt. Same with noise, or anything motion compensated.

    *The actual speed is 30000/1001 and 24000/1001
    *2 I should mention that using repeat field flags, the repeated fields are perfectly repeated (the current video field is just read out again). If you happen to have repeated fields in the encoding, the compression is lossy, the repeated field is not exactly the same. It's another reason to mess up the IVTC by the TV etc.
    Last edited by jmac698; 13th Nov 2011 at 04:42.
    Quote Quote  
  17. Originally Posted by manono View Post
    Originally Posted by sanlyn View Post
    I mean, what sort of signature would differentiate this between PAL->NTSC and FILM->NTSC?
    jagabo already mentioned the difference. The former could be 3:2:3:2:2 and the latter 3:2. I find the chances of the former remote, as I have yet to see a retail DVD sourced from PAL and converted to NTSC like that
    I've seen it once or twice. Even on cable TV. But, like you, I've seen a lot more PAL to NTSC conversions with field blending.

    Straight 3:2:3:2:2 pulldown can be removed with TFM().TDecimate(rate=25), but field blended material needs to use something like Yadif(mode=1).SRestore().
    Quote Quote  
  18. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Thanks again, jagabo and everyone, for clarifying. I'll soon be geared up to try just about anything.

    Now, some of this comes as news. Since my #1 visit to this forum, Doom9, and digitalfaq 7 years ago the rule was that interlaced video targeted for DVD/TV should never be deinterlaced, whether telecined or not. I don't doubt anything I'm learning here, but after staying up and reading stuff on the 'net til 2-AM (I really have to stop that) I get conflicting instructions. Progressive isn't really progressive, interlaced isn't interlaced but it really is if it's A or C and not B, etc. When you say TIVTC doesn't deinterlace but restores the original progressive sequence (which is really interlaced by the time it gets to DVD, because "standard DVD is inherently interlaced" ), then I have to twist my thinking and methods about rendering MPEG. I also see warnings where cut-edit with 24fps progressive AVI has certain hazards to be avoided, and that cut-edit the resulting MPEG should be done only on certain frames, and that GOP grouping for I, B, and P frames should be altered for this-or-that-situation. Don't cut out the last P frame in a group - even though TMPGenc MPEG Editor seems not to mind, and re-renders the joined groups to look correct if they break GOP rules. I prefer to avoid re-rendering MPEG, even if TMPGenc does do a pretty clean job of it. I've followed these GOP and cut-edit cautions for a long time.

    I'm seeing "never deinterlace an interlaced video" -vs - "don't interlace progressive video" -vs- "the NTSC and PAL standard for DVD/TV is always interlaced." My interpretation of "progressive" (whether from film source or not) is that progressive is not physically interlaced. So the question is, does TIVTC deinterlace as well as de-telecine and decimate? (My answer: if I play the IVTC I created without using SeparateFields(), Jackie's yellow dot appears on two frames. If I use SeparateFields(), the dot appears 3 times. Doesn't this tell me the IVTC'd clip is interlaced?).

    According to the DVD FAQ (http://dvddemystified.com/dvdfaq.html#3.4):

    525/60 (NTSC, 29.97 interlaced frames/sec) and 625/50 (PAL/SECAM, 25 interlaced frames/sec) video display systems are expressly supported. Coded frame rates of 24 fps progressive from film, 25 fps interlaced from PAL video, and 29.97 fps interlaced from NTSC video are typical. MPEG-2 progressive_sequence is not allowed, but interlaced sequences can contain progressive pictures and progressive macroblocks. In the case of 24 fps source, the encoder embeds MPEG-2 repeat_first_field flags into the video stream to make the decoder either perform 2-3 pulldown for 60Hz NTSC displays (actually 59.94Hz) or 2-2 pulldown (with resulting 4% speedup) for 50Hz PAL/SECAM displays. In other words, the player doesn't "know" what the encoded rate is, it simply follows the MPEG-2 encoder's instructions to produce the predetermined display rate of 25 fps or 29.97 fps.
    So my short 24fps test for my LCD, plasma, and CRT and 4 different DVD test players will be:
    input: 24.976 fps interlaced, 640x480 (the titles recorded on DVD are 720x480).
    output: 24.976 fps with 3:2 pulldown for playback, interlaced, 720x480, 4:3 aspect ratio.
    For different scenes, bitrate = VBR 6000 target / 7500 max (6500 target for critical motion or high detail).

    My rendered output GOP is closed group, IBBPBBPBBPBBPBBP (16 frames per group, which is the way my DVD recorder groups frames, but I'm reading where (a) this isn't good grouping, and (b) this grouping is OK. I sometimes use a 13-frame IBBPBBPBBPBBP for fast-motion sequences.
    Quote Quote  
  19. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    So you guys battered my hard-headed resistance to change, and I'll make a new version. But I do have to keep a promise and complete the old one by Christmas -- to be replaced by a new one later (like, next Christmas!).

    Here are 7 seconds of unmodified VHS capture (YUY2 AVI, huffyuv v2.1.1-CCESP Patch v0.2.2, 640x480), 60-MB. Oughtta be enough motion and detail to ascertain a few things (looks like 2:3:2:3:2:3 to me). Ed: I remade this huff version because huff/Avisynth on some machines have trouble with huffy stuff downloaded from other PC's (I do sometimes, too):
    http://dc309.4shared.com/download/4oig22PF/LIL3_8662_C.avi
    For huffyuv haters or problems, here's a Lagarith YUY2 version of the same clip (54 MB):
    http://dc247.4shared.com/download/sT4BYxVU/LIL3_8662_CL1.avi
    or a 2-second Lagarith version (13MB): http://dc441.4shared.com/download/vKEBEJVP/LIL3_8662_CL2.avi

    If you want the screechy audio for this clip to play with, here 'tis:
    http://dc307.4shared.com/download/MxQQofKf/LIL3_8662_C.wav

    Herein is 1 minute of short, abbreviated clips from VHS, in MPEG2 format (AC3 audio). Except for YUY2->RGB conversion in Avisynth, the scenes are unmodified. Since the project began in 2007, some image problems were fixed, some weren't. As seen here, the colors ain't right, some YUV level adjustments should have been done first, there's some flutter, bad frames, spots, chroma and tape noise, burnt highlights, combing, and whatnot. These are from capture #3. There are 4 others, but I might have to make #6 before it's done.
    NTSC MPEG2, 720x480 (640x480 display) Dolby audio, 46-MB, 1-min 1-sec.
    http://dc439.4shared.com/download/iplFapbI/LIL3_CE.mpg
    Last edited by sanlyn; 13th Nov 2011 at 16:50.
    Quote Quote  
  20. I can't get the AVI file to decode properly:

    Name:  corrupt.jpg
Views: 47
Size:  111.1 KB

    The MPG file is normal 3:2 pulldown.
    Last edited by jagabo; 13th Nov 2011 at 12:30.
    Quote Quote  
  21. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Originally Posted by jagabo View Post
    I can't get the AVI file to decode properly:
    Hmm, didn't some of us have similar problems with huff avi's in some threads about Cher videos?
    http://forum.videohelp.com/threads/339411-Tweaking-VHS-captures?p=2110373&viewfull=1#post2110373

    FYI, ffdshow is completely disabled on my PC. I still think ffdshow on a machine has something to do with it, and I see many posts about people getting different results. All my huff avi's play OK in media players on all my PC's except in VLC player. If I have ffdshow's huff turned on, the opposite occurs: VLC player plays it, nothing else will.

    Rather than redirect you to the whole 60-MB, I made a 1/4 length (about 13MB) version of the above using Lagarith YUY2: http://dc441.4shared.com/download/vKEBEJVP/LIL3_8662_CL2.avi

    Also I used the same scene in the MPEG sample posted earlier at http://dc439.4shared.com/download/iplFapbI/LIL3_CE.mpg
    Last edited by sanlyn; 13th Nov 2011 at 14:11.
    Quote Quote  
  22. I've tried with HuffYUV 2.1.1, HuffYUV 2.1.1-CCESP Patch v0.2.5, ffdshow, and VirtualDub's internal HuffYUV decoder. All failed. VLC internal decoder can't play it properly either.

    In any case, the MPG video and the Lagarith video are normal 3:2 pulldown.
    Quote Quote  
  23. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Oh, brother! Sorry, jagabo, I must've hit the wrong button. But thanks for taking a look. I'll make the AVI again and replace the old one with new huff and Lagarith 60-MB versions.

    Have no idea what's up with huff on various PC's. I've been spotting problem posts here and on other forums for years occasionally. I shuttle huff avi's among 4 PC's in my own setup - even my own PC's have a huff problem now and then. No one seems to have the answer.
    Last edited by sanlyn; 13th Nov 2011 at 16:55.
    Quote Quote  
  24. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Just tried a 5-minute cut of IVTC'd / 3:2 pulldown material as suggested (er, as commanded, more or less). Didn't look good on the PC or TV -- combing and stutter on movement, soft image, grainy. So, I changed one setting in TMPGenc Plus 4:3 pulldown setup.

    TMPGenc Plus -> Setting -> Advanced -> Video Source Type: Non-interlace (progressive) NO
    TMPGenc Plus -> Setting -> Advanced -> Video Source Type: Interlace YES

    Very smooth playback now.
    Quote Quote  
  25. On the Advanced tab the Video Source Type should be "Non Interlaced (progressive)". Then on the Video tab:

    Name:  tmpg.png
Views: 261
Size:  31.4 KB

    The lagarith file IVTC and encode with TMPG as above attached.
    Attached Files
    Quote Quote  
  26. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Yep, those are my frame rate and encode mode settings. Previously I didn't think to check that "source type" setting. The MPG looks much cleaner than the "old" version I'm about to complete (let's just call the old one a trial run. Hm?). Thanks for your brief "preview" of the new version.

    God heavens, what am I getting into this time?
    Quote Quote  
  27. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Originally Posted by manono View Post
    Movies shown in movie theaters in PAL countries are shown at 24fps, just as they are everywhere in the world. Or maybe you talking about home video versions available in PAL countries? I can find no record of a DVD of the film ever being released anywhere. Maybe only VHS tapes are around.
    I see only VHS, and not that easy to find. Some years back when I first bought my VHS copies I noticed PAL versions at Amazon. I do recall seeing a laserdisc version playing as demo in a high-end a/v store in NY City, but waaay back, a long time ago. I remember the LD looked dim, soft, and grainy like the cable version.

    The film did "sort of OK" in the U.S., maybe because the title song became a hit. 5 Oscar nominations: Leslie Caron best actress, art direction, screenplay, cinematography, and won the Oscar for best musical score. I can see why the camera work and color were nominated. If you get the color back in shape from the VHS it's gorgeous just to look at. But the movie's a Paul Gallico fairy tale with a few musical interludes - not a formula for box-office boffo. Amazon still sells the VHS, at nearly 3X the old price.

    The movie was adapted as a hit Broadway musical some might remember: Carnival, with hit songs (Love Makes The World Go 'Round). Hit your party guests with that trivia and they'll all leave early.

    Thanks for helping to push me into re-doing this correctly. Now that I'm looking up info on IVTC, I see it all over the place. How did I miss it earlier? Gotta refine my Google search, I think.
    Last edited by sanlyn; 14th Nov 2011 at 08:18.
    Quote Quote  
  28. Member
    Join Date: Sep 2011
    Location: NTSC
    Search Comp PM
    I thought I'd look it up. Well, your troubles are solved. It's playing in HD Feb 19.
    http://www.tcm.com/tcmdb/title/81426/Lili/

    Do you have a QAM capture card? Well there's a couple ways to record this. If it's on cable, and there's a driver for the cable box, you can record through firewire. If it's on cable, and the channel is in the clear, you can record with a tuner card. You can use a Colossus or other card to record through HDMI or component.
    If it's not a new version, you'll still be fixing the spots, but it will still look better than your VHS copy.

    It's also available on 2 dvd's, here's one:
    http://www.amazon.com/Lili-NON-USA-FORMAT-PAL-Reg-2/dp/B004C65AM6/ref=sr_1_3?s=movies-...1286572&sr=1-3
    The other seems to be low quality, don't know about this one though.
    Last edited by jmac698; 14th Nov 2011 at 10:10.
    Quote Quote  
  29. Banned
    Join Date: Oct 2004
    Location: New York, US
    Search Comp PM
    Thanks, jmac698. I'll check the broadcast. But I get channels only thru my cable boxes. Most cable providers in the U.S. have made this mandatory (which means my $400 HD Samsung outboard tuner has been useless since September 2010, except for over-the-air - and no antenna in my co-op building in New York). An external QAM tuner won't pick up a thing thru that cable line. The local company's newer HD boxes even have all outputs disabled except HDMI and digital audio (I hate HDMI. Garbage).

    TCM broadcasts a low-bitrate print, very noisy. Their HD channel is copy-protected thru my HD box. I can record from that HD box via the s-video (480i only, letterboxed and pillared) or component outputs (1080i), but TCM and most other HD channels here are copy-protected. I still record TCM thru my older digital-SD box - but even then, many SD downsampled channels are copy-no-no. They record, but I can't burn to disc from my recorders (DVD-RAM transfer only). If the past is any guide, the VHS is more workable than the cable version.

    I'm familiar with the DVD. No PAL player here. Players available, but poor quality (one is infamous for rotten motion handling and chroma upsampling errors) - and my TVs are NTSC. But I've seen posts in the UK saying it's the same PQ as the 2nd-rate cable job. As for buying new gear -- that's out, now that I finally shot my funds by buying After Effects.

    I've recorded the TCM print. But I'll be tuned in, just in case they improved it. The PBS analog broadcast that I used here (picked up thru my Samsung tuner a while back) has the cleanest titles so far.

    P.S.: sometime after Christmas I'll try your ideas with monitor calibration. No time at all for now. (<- off-topic yet again?)
    Quote Quote  
  30. Member
    Join Date: Sep 2011
    Location: NTSC
    Search Comp PM
    older digital-SD box - but even then, many SD downsampled channels are copy-no-no. They record, but I can't burn to disc from my recorders (DVD-RAM transfer only)
    Hmm, let me guess that you're sayiing that your DVD recorder, which has a hard drive, lets you record the channels but refused to burn to disc. I've never used a DVD recorder, but I've never had that limitation. You can pick up a cheap USB TV Tuner for about $35, but I don't know if they respect Macrovision signals or not. None of my cards even notice Macrovision, so I can record anything from my cable box.

    Not only that, I get TCM unencrypted through my PC tuner card, with QAM. If that didn't work, I could use record through my firewire port of the HD cable box. It has some limitations; I can record any channel live but I can't record material saved on the DVR.

    Finally I could record through svideo directly to PC through my tuner card. I've also invented a technique to let any PC tuner card record 480i or 480P component video. I have the same limitation in that the 480i output isn't affected by Pan Scan/Widescreen settings of the cable box, but 480P is, so I can get anamorphic, widescreen, 480P, component video, with just a normal tuner card

    Finally if none of those work, there's a way to record it off the TV screen, but we don't want to go to those lengths

    Also, that PAL DVD, is not a problem at all. You can copy the files to your computer, and convert them to NTSC, then burn them back to DVD. Even if it's a cheap DVD which is a copy of the PAL VHS tape, the PAL tape could be better quality, for one thing PAL is 720x576 so there's more resolution, second because PAL color tends to be more stable (reportedly; not a lot of experience with this).
    Quote Quote  



Similar Threads