VideoHelp Forum




+ Reply to Thread
Results 1 to 26 of 26
  1. I have a Hi8 camcorder, and i'm trying to capture a wide screen scene, wich is streched in a 4:3 format.

    Lets assume that after transfering this video to the computer, it will be in 720x480.

    To restore the aspect ratio of the video i have to strech the picture to 853.3 x 480, so i will get the perfect aspect ratio of the picture.

    but this resolution is not possible, so i have to strech it more. In my case : 1280x720 ( 1.85:1 )

    By doing this i'm getting some ugly vertical lines, because i'm streching the horizontal resolution.

    So my question is :

    How should we deal with this kind of shots to preserve the resolution, when editing wide screen pictures ?

    Cause i'm using bilinear techniques and bicubic techniques, but vertical lines still appear.

    Quote Quote  
  2. Member vhelp's Avatar
    Join Date
    Mar 2001
    Location
    New York
    Search Comp PM
    hi LordVader,

    The correct way to AR a cam's 16:9 source (assuming you shot in 16:9 mode)
    is to..

    Once exported to your hd, and assuming 720x480, you would:

    * Video tab: aspectRatio: [16:9 Display]
    * Advanced tab: SourceAspectRatio: [16:9 525 line NTSC)]
    * Advanced tab: Video ArrangeMethod: [full screen] or [full screen (keep AR)]

    Note, the 2nd "full screen" will give you slightely smaller files, but might
    be a teeny-tiny bit off. Actually, I like cuttin goff 8 pixels from left/right since
    my TV set cuts those areas off anyways. So, it's your call.

    Since the video is allready 16:9 (ie, 720x360, but resized/stretched to the
    720x480) you don't have to mess w/ it during encoding. Just setup TMPG
    the way I have it laid out (above) and all should be well.

    Note, pretty much all cams only gives you the 1.77:1 AR output. I've never
    seen one that does 1.85:1 or 2.35:1 AR's.

    Now, when played on your normal 4:3 TV set, you'll the black bars.
    But, if you play it on a widescreen tv set, it will play fullscreen (no bars)

    -vhelp 2103
    Quote Quote  
  3. you're totally right.

    Actually, i'm shoting in 1.77:1 streched, and i've been doing my compressions to mpeg-2 with display mode in 16x9.

    that's because i beleive i can get more resolution in the image with this method.

    But like i said, i still don't know how can i get rid of this vertical lines.
    Quote Quote  
  4. Member FulciLives's Avatar
    Join Date
    May 2003
    Location
    Pittsburgh, PA in the USA
    Search Comp PM
    Why can't you just leave it as 720x480 16x9 while doing your editing. Yeah everything will look stretched if you are editing in a fixed 4:3 window but who cares? Surely it's good enough for you to see (despite the aspect being wrong) for editing purposes and when you make your final DVD (I assume that is the final target ... a DVD disc) just make sure you set it as 16x9 anamorphic in your authoring stage and it will play correctly on any TV be it 16x9 or 4:3

    - John "FulciLives" Coleman
    "The eyes are the first thing that you have to destroy ... because they have seen too many bad things" - Lucio Fulci
    EXPLORE THE FILMS OF LUCIO FULCI - THE MAESTRO OF GORE
    Quote Quote  
  5. Ok, just forget for a second, what i said about resizing the image.

    Lets assume that i'm compressing my 720x480 videos, directly to MPEG-2 with 16x9 display mode.

    It looks right in my TV, it looks in 16x9 aspect ratio. But that's not what i'm talking about. I'm not having problems with that.

    What i'm trying to fix, is the jaggy lines that appear vertically, when you shoot in 16x9 mode. Got it ? :P

    See the picture in the topic ?

    I know my english is bad, but not that much.
    Quote Quote  
  6. No we (at least I) can not see the picture in the topic it's just a red X. Could be interlacing could be artifacts, could be a lot of things. For the record, you said everything looks fine when played on your standalone? Or do you mean that the DAR is ok when played on your standalone but that these lines appear?

    Either way there's no need to resize, just capture in 720x480 and author an anamorphic DVD. I suspect that the lines are due to your interlace source. If you source is interlaced re-sizing the vertical resolution is not a good idea.
    Quote Quote  
  7. the lines are not horizontal ( i know what interlaced is )

    the lines are vertical ( they are not noticeable on the TV, but i can see them in the computer )

    for example, have you ever compresses a video in 352x480 ?

    what do you think that happend, when those 352pixels are streched in your TV, don't you see a little vertical lines ?


    Well, that's exactly the lines what i'm talking about.

    ---------------------

    They are not exactly "lines", they look like little up stairs in the borders of objects. It is caused when the picture does not have enough horizontal resolution. For example 352x480.

    352 is not enought resolution to compensate the vertical resolution of 480 pixels, right ?
    So... when 352 pixels are streched to fit your TV screen, a little imperfections appear vertically on the screen.

    I don't know how to explain, hope you understand what i'm talking about.
    Quote Quote  
  8. I'm a Super Moderator johns0's Avatar
    Join Date
    Jun 2002
    Location
    canada
    Search Comp PM
    I know those vertical lines you are talking about,i thought it was low bitrate at first when i started to do 352x480 but the lines stayed at higher resolutions so i just learned to ignore them.
    Quote Quote  
  9. i'm working for some projects that will be analized by some people in the digital industry, and i'm competing with a horrible hi8 camcorder.

    i just want to do my best.

    I've tried all resizing techniques availables, but i can't get rid of this lines.

    shooting anamorphic with black borders is insane. Small targets are totally unperceptibles. And if i try to diffusse this lines, It will look blurry in a big TV screen. I know those guys will not ignore that

    Quote Quote  
  10. Member vhelp's Avatar
    Join Date
    Mar 2001
    Location
    New York
    Search Comp PM
    Hi LordVader,

    Once again, I know what you're talking about!!

    This is a very common anamily w/ low-budget cams, and
    even my Sony TRV-22 suffers from this phenomina of what I have
    refered to as "saw-tooth" from shooting in 16:9 mode, of which
    has to do w/ the poor nature of the cams' built-in cropping
    and resizing in during footage.

    Read here (below) for a clearer understanding (so I don't dup
    licate what I've said already )

    * Shooting DV footage w/ "saw tooth" output issues

    If you want more info, come back, but first research what I've
    already brought to light.
    .
    .
    yeah, it's time we (you guys) waik up to those 16:9 modes and
    their issues.

    -vhelp 2104
    Quote Quote  
  11. Member FulciLives's Avatar
    Join Date
    May 2003
    Location
    Pittsburgh, PA in the USA
    Search Comp PM
    Originally Posted by vhelp
    Read here (below) for a clearer understanding (so I don't dup
    licate what I've said already )

    * Shooting DV footage w/ "saw tooth" output issues

    If you want more info, come back, but first research what I've
    already brought to light.
    Where can I find the 411 filters mentioned in the doom9 thread? I notice you said you use something called 411 Helper and that it is a VirtualDub filter but is there one like it for AviSynth?

    Can't seem to find these 411 filters anywhere.

    Thanks,

    John "FulciLives" Coleman
    "The eyes are the first thing that you have to destroy ... because they have seen too many bad things" - Lucio Fulci
    EXPLORE THE FILMS OF LUCIO FULCI - THE MAESTRO OF GORE
    Quote Quote  
  12. Member vhelp's Avatar
    Join Date
    Mar 2001
    Location
    New York
    Search Comp PM
    Hi FulciLives,

    Yeah, the 411 filter (it's effects) varies from source to source ie, when
    capturing from Satellite for instance, you can see a difference in the red
    and blues (where you notice "square blocks". when this 411 filter (speaking
    for vdubs' filter) is used, you can see a difference, and you can see
    the blocks disapear. It's not the greatest, but does work when I use
    my ADVC-100 w/ my Satellite ie, I captured some "Superman The Movie"
    and there were "square blocks" in the DV file.

    Let me get right back to you on those filters.

    FWIW, I don't really notice diff. when I use AVS filter.. but I didn't realy
    test it out much. So, who knows. But, I use the vdub filter and I know
    it works their.

    Be back in a minute.
    -vhelp 2106
    Quote Quote  
  13. Member vhelp's Avatar
    Join Date
    Mar 2001
    Location
    New York
    Search Comp PM
    Ok.. I'm back

    You can get the filter at doom9's trbarry's website for the AVS version, and
    just do a quick Ctrl+F search for "ReInterpolate411" and click download.


    * ReInterpolate411.zip


    And, for the vdub version, done by Xesdeeni, search for the fle
    name "411Helper.zip"

    * Xesdeeni's Web Page for --> "411Helper.zip"

    -vhelp 2107


    .
    .
    From VHELP's webpage, Before and AFTER, (note the blue monitor)
    Capture sample from my ADVC-100 w/ Satellite (s-video)

    Before, w/ 411Helper on:


    After, w/ 411Helper on:
    Quote Quote  
  14. Member FulciLives's Avatar
    Join Date
    May 2003
    Location
    Pittsburgh, PA in the USA
    Search Comp PM
    Hello Vhelp

    Thanks for the links and the images.

    Currently I capture with an analog TV tuner type card which allows me to use YUV 4:2:2 with either MJPEG compression (what I normally use) or sometimes I use HuffyUV (though usually only on short clips due to my HDD always being low on space).

    Anyways I've been very happy image quality wise with this setup up although I've got things setup so this rarely happens ... well ... I still do have audio sync problems from time-to-time and I really would love to get the Canopus ADVC-100 or DataVideo DAC-100 since they seem to convert/capture without sync problems.

    However I have been put off of using this solution due to the fact that DV uses 4:1:1 and the problems that come with that ALONG with the saw-tooth problem which I'm still not sure is a 4:1:1 problem or a DV codec problem.

    So it is nice to know if I do get the Canopus or DataVideo that there is a filter that more-or-less compensates for these "problems".

    As for AviSynth VS VirtualDub I have found that AviSynth is faster than frameserving from VirtualDub so that is how I mostly do things now (using AviSynth that is).

    - John "FulciLives" Coleman
    "The eyes are the first thing that you have to destroy ... because they have seen too many bad things" - Lucio Fulci
    EXPLORE THE FILMS OF LUCIO FULCI - THE MAESTRO OF GORE
    Quote Quote  
  15. Member vhelp's Avatar
    Join Date
    Mar 2001
    Location
    New York
    Search Comp PM
    Hay FulciLives,

    Canopus ADVC-100 or DataVideo DAC-100 since they seem to convert/capture without sync problems.
    That's one reason (or another one) why I got the ADVC-100. Naturally,
    since I don't have the DAC-100, I can't say for sure if it's "audio sync" 'less
    or not, ..and I don't want to soil it w/ bias

    As for AviSynth VS VirtualDub I have found that AviSynth is faster than frameserving from VirtualDub so that is how I mostly do things now (using AviSynth that is).
    If it's faster on your system (per your setup and things) then, that is all that
    matters, that you get faster throughput w/ your given setup.
    .
    .
    I have nother reasons for using vdub in the majority of my fun projects and things.
    .
    .
    And, another reason why I use vdub, is because, from what I have learned
    and understood, it's final output is RGB. Now, correct me if I'm wrong, but
    isn't TMPG RGB driven ?? That is, it prefers to receive RGB sources,
    becuase of it's internal structures and things. Well, anyways..
    .
    .
    FWIW, I do use AVS scripts in my mixup of things when I frameserve between
    vdub and AVS into TMPG. There are methods to MY maddness hehe :P

    Perhaps a little OT here. . . but. . .
    .
    .
    Oh, by the way, speaking of pics and things.. did you notice anything diff about that
    Superman pic I posted ??
    After a careful ananlyisis, I've concluded (based on your past inputs and DJRumpy's AR guide)
    I found its AR to be 2.35:1 though a little off by a few pixels. But, that got
    me confused because the source was 4:3 (as captured by my ADVC) and the video is quite
    punee (very thin) After cutting off 104 pixels (top/bottom) I got it to 272 (w/ some
    black bars left in for 2.35:1 reasons) But, it didn't look right either way when I got
    finish w/ the test encode.
    .
    .
    Well, I though I'd would just mention this. If you have any comments on this issue
    (though strange to me) I would appreciate it. Hay, I noticed that "Blade Runner" when
    I capture it from SCI-FY too, it's like this. Anyways...

    -vhelp 2108
    Quote Quote  
  16. I am interested in this problem myself. I have not used my cam in 16:9, but is did not cost $10k + so I will assume I will have this trouble. Actually, I have seen this in 4:3 on white rails in front of a dark background, where the shot was at a slight angle.

    I think there are 2 problems.

    #1 The NTSC 4:1:1 problem where the color is shared by 4 pixels so it bleeds out. In VHelps example, you can see more of a problem in the red lights to the right. They a blocky before, but smooth after. The upsize, smooth, downsize seems to help a lot here. It's hard to believe a good DV codec can't just do this for us.

    #2 jaggies are an artifact of CCDs. I'm not sure why. Here is a discussion of the topic by some very knowlagable people. It about 16:9.

    http://www.slashcam.com/advlfaq/516.html

    I must say, I do not understand where Lordvader 1st sees the problem. After inital transfer to PC? After resize to 1280x720? I think it is there before any resize, but the resize makes it bigger. I'm guessing the red lines were added to just show the problem.


    PS: The picture posted by Lordvader can be seen by right clicking, copy the url, past it into a new browser window. His provider does not allow external site links. He may not notice because his browser has cached the image. This has happened to me too.
    Quote Quote  
  17. I'm a Super Moderator johns0's Avatar
    Join Date
    Jun 2002
    Location
    canada
    Search Comp PM
    Will those filters help out the jagged lines i get when i reencode my 480x480 svcd backups to 352x480 dvd compliant?
    Quote Quote  
  18. Member vhelp's Avatar
    Join Date
    Mar 2001
    Location
    New York
    Search Comp PM
    .
    .

    Lets talk about jaggies or saw-tooth for a moment...

    My belief (based on many tests and observations) for the cause/reason of this
    phenomina is due to the Interlace setup in a CAM. Remember, those are fields that
    are being recorded, NOT frames. If they were frames, your source would be FRAME-SIZE
    ie, 720 x 480 !!
    .
    .
    The Celeste Challenge
    But, because your cam is interlaced, it is actually capturing (taking snapshots) of
    720 x 240. Guess what, those harizontal issues you all (include me) are
    speaking of, this is why. becuase our height is only half, and as such, is only
    capturing half the detail, hence our horizatal artifacts or saw-tooth/jaggies.
    This is the cause - plain and simple. You can test for yourself this fact, by
    observing your footage while connected to your TV set. Run the Celeste box across
    the cam, and at various arches etc. Look closely at the horizontals and see how
    limited they are in detail. Notice how they begin to saw-tooth depending upon
    how you angle/arch your box.
    If you cam was full-frame, it would capture it as 720 x 480, and you would NOT
    see those horizontal jaggies. Plain and simple.
    .
    .
    That's your reason/cause !!


    Next issue. . .
    Also, I too, have done a lot of testing of these issues, in my quest for a better
    (or best) end product from DV cams.

    When taking footage w/ my TRV-22 cam in 16:9 mode, there are horizontal jaggies and
    so on (read above) . There are also some color artifictcs as I'm expressed over at
    doom9 (see link in my prev. discussion above)
    .
    .
    After toying around w/ 16:9 vs. 4:3 mode footaging for some time now, I have concluded
    that for best and maximum quality from DV source, it's best to shoot in 4:3 mode and
    then later, crop the top/bottom 60 pixels off (each) and encode to 16:9 (1.777) and enjoy.
    .
    .
    I go a step further, and enocode to a 2.35:1 instead because the quality is kicked up
    a notch or too. And, couple that w/ some night-time footaging, and you can see the end
    results, as in my sample clip I posted (see link above - VHELP's Samples.. )
    And, that clip was even de-interlaced (vhelp's way) although, I think I have improved
    on my de-interlace techniques since that clip first aired
    .
    .
    But, my method of producing a 2.35:1 project does not actually envolve an AVS script
    and it's related filters. The majority of my DV projects entail vdub and TMPG. I guess
    you could blame that on the WYSIWYG'ness of vdub and it's scripts or even AVS scripts
    for when I DO feed inside vdub. I never feed an AVS script directly to TMPG.
    It's always a vdub (*.vdr) server.

    As far as loosing resolution in going this route. Well, that's a matter of opinion.
    I feel, that because I am the director of my footage, my aim is 2.35:1 output.
    I'm still testing/debuggin a wide-angle lens I got. This lens really adds more body
    or area. So, the way I figure, is like this. If I use this WA lens and incorporate
    my 2.35:1 process, and depending on the way you look at it (opinion) we'd get the
    full resolution. But, as I said earlire, that's a matter of opinion


    Getting back on topic. . .

    So, to rule out (or minimize) as much as possible, those artifacts:
    * It's wiser to shoot footage in 4:3 mode
    * Then just crop off the top and bottom by 60 pizels each
    * Making your final source 720 x 360 from this cropping
    * Then, resize to 720 x 480 and
    * frameserve into TMPG and encode w/ 16:9 as your aspect ratio.

    In going this route, you can only gain in quality w/out those stairs, as in my
    sample clip as proof of pie (or is it pudding) ??

    Do give the above some thought.

    From the Video Workstation of,
    -vhelp 2110
    Quote Quote  
  19. Hi guys, happy to be here again since yesterday.

    In fact the Red lines, was only to show where are the jaggy lines, that's all.

    I'm going to read all the replys very carefully. I've found very interesting the thing about 411 filters. I'll do my new captures.
    Quote Quote  
  20. Originally Posted by vhelp

    Lets talk about jaggies or saw-tooth for a moment...

    My belief (based on many tests and observations) for the cause/reason of this
    phenomina is due to the Interlace setup in a CAM. Remember, those are fields that
    are being recorded, NOT frames. If they were frames, your source would be FRAME-SIZE
    ie, 720 x 480 !!
    I think you are dead on as to the stair step problem. I do not have enough experience to have this myself without doubts. An interesting conclusion one can draw from this is that when a cam/DV device does an analog conversion/sample, this will NOT be a problem. Well maybe, depending on how it is done, but assuming they have a chip/process as good as those cheap capture cards, this should not be a problem.

    This is why I think it is due to the limited sample you get from a CCD.

    I am pondering your other statements.
    Quote Quote  
  21. OK, after making some new tests. I've solved the problem by separating the fields and putting them in different layers in after effects.

    by moving the first layer a little to the right ( with 50% opacity ), the frame fit exactly on the other layer and the lines dissapear. Then, saving as avi and resizing to 1280x720 with "precise bicubic 1.00" i get rid of this lines completely.

    -----------------------

    of course, i can't do this process with my home videos, cause it needs to be recompressed, and i don't have to much disk space.

    It's a very ridiculous process to get rid of this lines.

    I refuse to shot in 4:3 and then cropping 60 pixels and encoding 16:9 display mode.

    The objects are totally unperceptibles, the image is incredible blurry. It could be done with great lenses, and i don't really have money to buy the 240$ ones that help me to get more depth.
    ...


    If any of you know a way to separate fields in diferents layers and apply opacity to the frist one, in virtual dub, that would be very nice.
    Quote Quote  
  22. @LordVader

    Great that you fixed your problem. Sounds like you deinterlaced,



    @VHelp

    Do you think the sawtooth problem is only a problem with 'filming' vs capture? If this is the case, the ADVC would not have this problem. Only the DV 4:1:1 NTSC color problem.

    Trev
    Quote Quote  
  23. Member vhelp's Avatar
    Join Date
    Mar 2001
    Location
    New York
    Search Comp PM
    Hi trevlac,

    Do you think the sawtooth problem is only a problem with 'filming' vs capture? If this is the case, the ADVC would not have this problem. Only the DV 4:1:1 NTSC color problem.
    Yes, Filming via a DV CAM is the cause. Not the ADVC-100 device.

    The "saw-tooth" effect is due to numerous nonsense of the DV cam(s) out
    there.
    .
    .
    Remember the KEY element I mention above ??
    720 x 240 is the main cause of this effect. The others that contribute the various
    artifacts are results of those cams and their respective weeknesses. Some examples
    are:

    * poor lens
    * poor resizing (weather in chip or circuitry)
    * the nature of the "process" ie, 16:9 mode and it's limits w/ 720 x 240 fields
    ...and resizing or merging (aka, Interlacing) source frames.
    * poor lighting conditions. ie, the lower the light levels, the harder it makes the
    ...cam device react to "reproducing" the source respectfully
    * much more other nonsense.

    However, if you were to hook up your cam to an S-Video can analog capture it, you
    would still suffer from the 'saw-tooth' effects. Sorry, but again, thats the nature
    of the lens and (720 x 240) fields being Interlaced into one frame.

    Just to note, (I failed in other threads) ...
    When the x240 fields are Interlaced into x480 frames, the effects of "saw-tooth"
    is not as noticable, though still their, giving you the illusion of sharpness.
    You won't notice this so much on a TV set (looks soo smoth from a distance) but
    when you capture it and view it through say, vdub, you'll notice it right away,
    unless you're a newbie to all this.
    .
    .
    ** I just finished posting on doom9 on a similar issue, under virtualdub/mod area.
    ** Poll: (Resize and Deinterlace) or (Deinterlace and Resize)?
    .
    .
    I think that in "some" cases, you can get away w/ a deI (deInterlace) w/ proper
    technique and a good eye, of which takes time to master. The clip I did yields
    pretty good results of this (still on my VHELP's Samples.. thread - link above)
    At least, I give good illusions of full frame encode from my DV footages hehe..

    From the Video Workstation of,
    -vhelp
    Quote Quote  
  24. Have you ever saw the picture quality of 1 field at a time ?

    the lines are more defined, and streching the picture to any size with "1.77:1" aspect ratio, with bilinear o bicubic mode, will smooth a little this lines.

    After making some tests, i got to say, that there is still something that not convice me at all. But it is fixable, making a little sacrifice with the resolution.

    With a little median, film glow, and motion blur, i get pretty decent results.

    The fields are not in the same screen area, i get stressed working with fields in different layers.

    If "saw-tooth" is a common DV cam problem, it doesn't make any sense that this thing does not happend in some widescreen TVs with S-VIDEO jack.

    Capturing process are so much different to video transfer processes, and i could deal with saw-tooth if it's something that comes directly from the tape in the transfer process.

    But capturing ? come on ?

    analog capturing processes, change video quality source at all. I don't see any reason of why something like this, is coming trought my TV card.
    How could know my TV card if video source is 4:1:1 or 4:2:2 ?
    my TV card is just capturing video.

    Is there any report where professional of the industry, talks about this problem, and doing nothing to even try to fix this problems in the next camera models ? of course not.

    Because i guess, they know how do deal with something like that, and what to do.

    And that's what we have to find.
    Quote Quote  
  25. Member FulciLives's Avatar
    Join Date
    May 2003
    Location
    Pittsburgh, PA in the USA
    Search Comp PM
    Well I only own an old Hi8 camcorder and never worked with DV or Digital8 but I thought I have read that there are special anamorphic lenses that you can buy for a camcorder that only records in 4:3 to achieve the 16x9 look ... the same way anamorphic 2.35:1 film is "squeezed" into a frame of 4:3 film.

    Since the camcorder would be in 4:3 mode with the lense doing the anamorphic stretching of the image then wouldn't this solve the saw-tooth problem if it really is the way the cam resizes the image?

    Just a thought ...

    - John "FulciLives" Coleman
    "The eyes are the first thing that you have to destroy ... because they have seen too many bad things" - Lucio Fulci
    EXPLORE THE FILMS OF LUCIO FULCI - THE MAESTRO OF GORE
    Quote Quote  
  26. with one of those lenses, i'll know what's really going on.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!