1. Join Now

    AVForums.com uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

50p vs 50i+deinterlacing+slowdown

Discussion in 'Camcorders, Action Cams & Video Editing Forum' started by Mandarin Man, Sep 5, 2005.

  1. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Hi,

    My first post on this excellent forum..

    I'd like to know what the added value of shooting in 25p is compared to shooting in 50i (PAL) using a low-end DV camcorder and then deinterlacing using software. You'd have 25 progressive frames in both cases. I know these two methods probably won't yield the same results, but I don't quite understand why not. Can anyone shed some light on this?

    The difference between progressively captured video and interlaced captured video is hard for me to grasp, since both are stored on tape as interlaced video, right? So would you get different types of interlaced video?

    Thanks.
     
  2. Roy Mallard

    Roy Mallard
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Ok, just about every conventional tv set and every standard broadcast is done in interlace.

    All of the odd lines are shown first (in consequetive order) followed by a scan of all of the even lines in the second field. These two fields form one frame on the tape, although they are actually scanned at very fractionally different times.

    Progressive scan playback is done by playing all lines in one field and in a logical consequetive order (i.e. no difference between odd and even lines)

    Because most camcorders are designed for viewing on televison rather than on a VDU, most camcorders capture in the 50 field interlace format.

    This creates a couple of issues:

    When you viewback footage on your editing PC sometimes there are 'jaggies' around moving subjects. This is because the PC monitor is displaying interlaced sourced fields in a progressive order. Viewing on a TV set will be fine.

    HOWEVER, if you are doing content intended for playback through a computer (say quicktime or windows media clips for powerpoint or web) then you would probably want to DEINTERLACE (make progressive) your footage to suit the monitor better.

    Some people want to pretend that they are shooting on film, and so believe that the motion artefacts of a 25fps progressive scan looks more like film than interlace (although neither is exact, with film you see all the frame at the same time, with interlace and progressive you only see a small area at the one time)

    Deinterlacing the footage can make pans tilts and zooms look absolutley terrible, really staggered.

    In actual fact 50i has far superior motion than 25p. There are a minority of users who will want to project onto a large screen, so in that instance taking 25p footage and printing it onto 35mm will work slightly better (though 35mm film is actually 24 frames per second)

    Most cameras record in 50i, a few offer progressive scan, Canon XM2, Panasonic GS400 (using a yukky pixel shift system so you only get 75% of potential resolution)
    Sony VX2100 and PD10x (but at 15 frames per second, full resolution but very staggered motion) and the XL2 and panasonic DVX100 & 200 (full res, full motion) and then most pro cameras.

    I shoot everything interlace, why?, because I don't shoot cinema, I shoot televison.
    If I was shooting cinema then I'd be shooting on super16 or 35mm.
    You can deinterlace footage at the production stage (and at least this way you have the choice of 25p or 50i, if you originate at 25p you cannot undo it) although this has its own set of issues (ghosting & motion trails around moving subjects)

    Unless you have a huge budget then there is IMHO no need to kid on you are shooting on film, even at 25p its apaprant that you are not (film has far more contrast latitude, shallower depth of field, more fluid grain artefacts) in my experience most folk who shoot on 25p are wannabe drama directors who have prosumer kit and think of themsleves as a bit 'geurilla', nothing wrong with that at all. but where theres no budget for film stock, there usually isn't any budget for a lighting director, a sound crew, decent actors or continuity blah blah blah. I think it is a gimmick to try to make otherwise crap footage look a wee bit special.
    Cinematic lighting at 50i will look far better.

    Now I know that this will really noise up some posters, but hey, it's all an opinion, and they will quote Mike Figgis and they will quote Andrew MacDonald, but I never listened to them in the past, I'm not going to start now, Ken Loach shoots on Film, Scorcese shoots on film, Adam & Joe shoot on MiniDV. Go figure.
     
  3. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Thanks for the elaborate reply, Roy.

    I think I have also found the answer to my last question:

    Of course, in the case of pogressive shooting, the data stream written to tape still interleaves the even and odd fields for proper interlaced TV display; it's just that both fields have been captured simultaneously instead of in even-odd alternation when shooting in 50i. This even-odd alternation is the reason for all the deinterlacing artifacts, right? So shooting progressive would eliminate these problems, if it's progressive video you want.

    BTW, is it possible to interlace 25p material for DVD authoring?
     
  4. Roy Mallard

    Roy Mallard
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    theres no point interlacing 25p footage, as you've correctly surmised the point of having two fields is that each field is slightly different (and to save on analogue broadcast bandwidth)

    If you deinterlaced a 25p frame all you get is two half res frames exactly the same, as the capture was 25p you can't really undo this, this is why I shoot 50i, you have some choice.
     
  5. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    If you are referring to the sort of footage you'll get from the canon and panasonics then its a moot point. The material is still intrinsically interlaced just that each consecutive field was captured at the same time as the previous one. Put them together in the right field order ( dominance) and you get a coherent frame rather than two mismatched fields.

    I'll be looking into the issues associated with encoding this sort of material to mpeg2 but don't think there will be any significant ones. Cutting out after odd fields to maintain the frame structure and flagging the dvd correctly during authoring are about it.

    As to the footage I actually quite like it. I shot some frame based footage on an XM1 I picked up last week for a couple of hundred quid and as long as you have the shutter wide open to minimise strobing and don't let the whites blow out too much the footage looks nice to me . Not much different to a film camera , if you have a smalll shutter angle you get strobing ( Gladiator , Saving Private Ryan etc). The frame based artifacts are no more objectionable to the interlaced ones ...just different.

    Prefer it to interlaced to be honest. I've messed around with a couple of optical flow based deinterlacing apps and whilst some were more effective than others I wasn't really blown away by the results compared with the native frame mode on the canon. The deinterlacers had their fair share of artifacts and smearing.

    A mate of mine has just bought the the cheaper sony HDV cam and raved about the quality of the footage when he deinterlaced it. When I took a look he was only dropping the odd fields and resizing the remaining one to PAL non-square pixel resolutions. It looked ok but it wasn't any better than the xm1 footage that we could see ( 16x9 resolutions excepted) and the deinterlace and resize process wasn't realtime ( it wasn't that slow but you wouldn't want to run hours of footage through it unless you had to).

    If you were careful with your exposure , and didn't kick up too much noise I see no reason why it wouldn't be a viable alternative to shooting on 16mm ( which I find well over-rated personally).
     
  6. Roy Mallard

    Roy Mallard
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Film is better for a few reasons, contrast ratio is approx 125:1 as opposed to 40:1 for CCDs (so better colour depth, more detail in dark & light areas) if you use high speed film for low light then there a far less obvious noise artefacts. Film cameras (generally) have better lenses than video cameras, because the image area is bigger you can do more with slective focusing.

    As a creative point, because film is more expensive there tends ot be far more pre-production done, I know this isn't a technical format issue as such, but it's one of the reasons why the film experience is different to the video experience, whether yer making them or watching them. But it's all a matter of opinion.
     
  7. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    Only if you are talking about negative densities.

    Most film negative is overexposed slightly to push the darker regions above the "toe" of the response curve to better record black detail in a more sensitive part of the film .

    This necessitates a print down on transfer to print film to perceptually increase the "contrast". In combination with the additional headroom that negative captures , the majority of which does not make it onto the "high contrast " print stock is not quite as hugely beyond video as you'd think. Especially if not every shot is nominally exposed negative vs print.

    The print range is still larger than video ( and this is before you bring colour gammut into the equation) but its not as massive as comparing a nominal negative range to 8 bit video, probably about half as large in the real world.

    And poorly exposed film is not going to look significantly better than poorly exposed video and most of the low budget films I see are poorly exposed generally ( some of the higher budgeted ones are pretty mediocre as well).

    Badly exposed 16mm is as noisy as badly exposed video . Super16 has significant stability issues as well.

    Saying all that if I was going out to shoot a low budget film , I'd likely have a crack at it using high def rather than super16 these days.
     
  8. Roy Mallard

    Roy Mallard
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Ok, this is now getting off the point of the post, but apply your printing degradation issues to reversal stock....

    And whilst you mention HDV, do you really think that there is no degredation when it jumps between MPEG2, the system codec then back to MPEG? That said, I think HDV has some potential at challanging 16, especially on sony HAD chips.

    Yes film is less salvagable than video if the exposure is out, but then it's back to best practice. I would ask why so much of what you've seen is out, is it a format problem or a practioner problem?

    By the same token how much tv is now shot on miniDV handycams than say 15 years ago?, would you say that production values have got better or worse?

    This is all off the point, we can argue about gamma curves, gamut and negative filtration till the cows come home, I maintian that home users have no need to shoot on 25p, and i would further attest that most of the intermediate budget video stuff that is being shot on 25p doesn't need it either.

    I think it is a great tool for very specific applications, but it's now so overused it's a cliche. It's main use has became vanity rather than to aid the viewing experience. "I want to pretend I shot this on film and kid on that it's really big budget even though it'll look crap on a tv set" (which is where 99.999999% of the footage will be seen) If folk are that bothered about being seen to shoot on DV, then maybe they shouldn't bother.
     
  9. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    The common practise of printing down pushed motion picture film is not a degradation its a useful way of increasing detail near black. The headroom in the negative stock makes this possible but there is not point including the headroom in an estimation of the available range vs video as its not transferred to print in its entirety ....so the difference in range between video and film is not quite as huge as is made out as the headroom on film accounts for about 1/3 of the available range which will not appear on print unless its printed down...was my point.

    I didn't mention HDV in comparisson to film I mentioned it in comparisson to footage off an XM1 in frame mode. However I agree it can challenge 16mm but I also think 16mm is over-rated and decent sd footage can rival it under the right circumstances anyway. MPEG2 simply put tries to hide the compression in motion where its hardest to notice , single frames might look a bit odd on certain footage but on the fly I find mpeg2 artifacts can be quite unobjectionable. there might even be compression advantages to using frame based material over interlaced with mpeg2 .

    Film is not less "salvagable" than video from an exposure point of view I don't know what gave you the idea I said it was. Quite the reverse in fact , video does not have much scope beyond what was intially exposed n the camera. Film can be printed down about 2.5 stops without objectionable clipping or crushing.

    My criticism was of the aesthetic quality of lighting on many large budget films I've worked on ( that I've often had to go to extreme lengths to manually relight in post ). Not a specific criticism of them having been exposed incorrectly.

    As for being off topic... I like the frame mode on the canons and I like the frame mode on the panasonics. I prefer it to interlaced and I feel that its a nicer look whether its displayed on a progressive display or interlaced. This might be because I find it more similar to film and prefer the way film looks over interlaced video . I prefer it to deinterlaced 50i footage and its useful format to shoot in if you want to do additional effects or post-processing.
     
  10. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    To be honest though I doubt the hardware mpeg2 encoder on a 3k camera is really going to be all that much cop. mpeg2 doesn't work best as a realtime encoder anyway.

    I was up in Scotland last week and the Scottish news coverage seemed to be using a fair bit of miniDV and that terrible newish Scottish soap opera looked a bit odd too ( some sort of half assed bobbed film effect I think).

    Any ideas Roy?
     
  11. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    The discussion of the the different shooting modes and cameras is beyond me as I'm not a filmmaker. I'm actually a film composer familiarizing himself with digital video. It's getting a bit out of hand though.

    Is this correctly put?:

    progressively shot DV -> intrinsically interlaced -> no deinterlacing artifacts (2 fields can simply be combined into 1 full res progressive frame)

    interlaced shot DV -> truly interlaced -> deinterlacing artifacts

    Also, if you have a 25p NLE project and export/compress this to say, quicktime (MPEG4) or WMV, the encoder will deinterlace the footage, creating a 25p file, right? Do these computer video formats only support progressive video?

    Thanks,

    Roy (yep, another one)
     
  12. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    yes

    yes even the best deinterlacers leave artifacts

    No they don't need to deinterlace. Its usually better to deinterlace on playback rather than turn the footage into frames.

    Some of the cheaper bundled apps that come with cameras and other bits of hardware might well do and not give you the option of specifying to deinterlace or not... and all deinterlacing techniques are not equal in quality.
     
  13. Roy Mallard

    Roy Mallard
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    River City is done on digibeta as far as I am aware, I think it's Sony 950w's, but don't quote me. And yes that horrible flicker is there, illustrating my point that 25p does not work on present day broadcast. It might suit the convention of the big screen better, but definately not for telly. My point was and remains: if the end use is telly then shoot in the best format for telly, which is 50i, no discussion at all. If the end result is for projection then 25p (as most LCD projectors are progressive). If the end result is cinema the shoot on film.

    My objection tot he xm1 (& other 3ccd panasonics and canons before the XL2 & AVX100) is that to get the progressive effect then you lose about a quarter of the resolution through pixel shift.

    If you don;t want the expense then go and do something else. Horses for courses.
     
  14. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Ok, so what about DVDs encoded from film sources? 25 progressive (film)frames are encoded to interlaced field pairs (2:2 pulldown). These 2 fields can in turn simply be combined into 1 full resolution progressive frame by a progressive DVD player without the need for complicated deinterlacing methods?, right? So I would expect no deinterlacing artifacts??
     
  15. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    Got it in one. The optimum deinterlacing for frame originated material is a weave ( puts the two fields together). Sometimes its even referred to as "no deinterlacing" . The only problem comes if the cadence breaks ...that is you cut on an odd field...this wrecks the field dominance and you'll get the wrong fields being combined until you get another bad edit and then it will all pop back together.

    And in reference to your original question this sort of footage is essentially the same as video shot on those cameras that offer a real frame capture mode.

    The strobing results from insufficient motionblur with small shutter angles ( or ccd capture rates) in exactly the same way that you get a strobey effect in the battle scenes in Gladiator and Saving Private Ryan. If you watch Collateral you can see the opposite effect where they needed a large shutter angle to let enough light into the digital cameras in dark scenes and it results in too much motion blur and a smearing effect.

    This is what bothers Roy but if you lock the shutter angle to give an appropriate amount of pleasing blur and modify your exposure using other methods (lights , ND filtering, gain) you should avoid the strobing...unless you like it ..after all Spielberg and Scott thought it was useful for certain things.
     
  16. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Aha, so this simple re-interleaving is actually what a DVD player's film mode deinterlacer does? Just a matter of putting the right fields together. And now I also know what 2:2 pulldown detection means: I already understood 2:3 pulldown and I thought 2:2 pulldown detection just meant deinterlacing, but it's actually the player's ability to detect frame based encoded material and apply film mode deinterlacing to it, right? (no bobbing, Motion-Adaptive Deinterlacing, etc.). Things are starting to make sense, now..

    I guess 1080i HDTV broadcasts of movies are encoded in a similar way (also frame based encoding to field pairs, and containing 2:3 pulldown flags in the US), can you confirm this?

    Thanks a lot for the info so far, Keith. I'm almost done.. :)

    - Could you give me examples of truly progressive material? 720p HDTV maybe or WMV HD? How about the 24 fps Quicktime trailers, true progressive or still inherently interlaced?

    - What's the best software to get detailed technical info on digital video files (codec, framerate, interlaced or progressive, etc.)?

    Thanks,
    Roy
     
  17. Roy Mallard

    Roy Mallard
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    both premiere elements and final cut express have a useful level of presets and custom control, obviously which depends on your platform. Premiere permits WMV and Quicktime (a lot of folk use Real, I'll check, but I think you only get real codecs on Adobe Media Encoder, a more advanced package bundled with Prom Pro) Pinnacle allows WMV and Real encoding but with less control and no QT (on studio at least)
     
  18. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    You are not going to see any fundamental difference between frame originated material thats contained either as interlaced or progressive in a given file system as long as both are displayed in an optimum way. The interlacing is beside the point in the case of this material.

    Material on domestic dvds is usually in fields.In the case of 3:2 flagging its there to aid compression not striclty to enable inverse telecine (correct deinterlacing). Its not the case that there are 24 discrete frames on a dvd and the flagging controls the interlacing : its interlaced footage that then requires deinterlacing after decoding. The masters most dvds originate from are interlaced D1.

    720p is essentially a replacement for interlaced video ( its used in situations where normally they would have used a field based video camera...sports , news, ENG, studio ). Film based footage is primarily sent out as 1080i as to all intents and purposes its easy to deinterlace with no downsides but requires less bandwidth than broadcasting it progressivley. Whilst field originated material is impossible to deinterlace "perfectly" compared with capturing as frames in the first place.

    Thats the idea , in practice quite a bit of film material gets broadcast at 720p as well with repeated frames to get the frame range up to 30p in some cases.
     
  19. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Thanks, Keith.

    And the above is correct, isn't it?

    BTW, is the D1 master used in DVD encoding SD?

    I'll stop bothering you with my inquisitiveness, now.. :)
     
  20. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    Essentially yes thats 2:2 pulldown ( which isn't really pulldown at all ) and the technical term for the correct deinterlace is a "weave" which essentially means just putting the two fields together which essentially is the simplest deinterlace you can get. The only poblem can be if there is an illegal edit which leaves a single feild hanging about which will break the deinterlacing. If the deinterlacer is ok it will detect this and drop to a bob deinterlace to ensure you don't get mismatched field artifacts ... (if its a bit better it will swap the field dominance first and see if that fixes things before going to a bob). Another bad edit will put the field domaninace back to where it was before and hopefully the deinterlacer will spot this and return to a weave.

    I find very very few illegal edits on most commercial dvds these days. ( it would be nice if the after effects jockeys realised the menus and trailers on dvds should really be frame based as well though)

    2:2 pulldown structure is actually harder to detect than 3:2 pulldown from a deinterlacing point of view . However I suspect this is more the case of deinterlacers being written primarily from a US centric perspective. if you have manual control of the deinterlacing type you can always set it by eye.

    Yes the D1 masters many dvds originate from are interlaced and are SD.
     
  21. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    Incidentally all motion adaptive means is that the deinterlacer will use a weave if the footage is primarly static and drop to a bob ( most likely) if it detects movement in the fields. ( this is primarly for field based materia0.

    If it detects a 2:2 pulldown structure it should stay in weave regardless.


    A bob effectively takes each field and resizes it to full height and then displays each new "frame" ( 50fps for PAL , when it deinterlaces correctly it repeats the frames however many times it needs to reach the refresh rate 50fps ...each frame shown twice). Switching from a weave to bob menas that you get a sharper picture on static shots and the softer bobbed frames are hidden by motionblur in the moving shots. Dscaler is a good example of this and it also has very good 2:2 and 3:2 detection routines.

    There are perpixel motion adaptive deinterlacers which essentially segment the image and perform a vector driven warp to reconstruct frames using both mismatched fields but this in itself softens the image and introduces artifacts. Its unlikely you'll find a realtime deinterlacer that manages this in a reasonable price range although there are software deinterlacers that use this technique albeit not in realtime . They have to render the footage.

    This is why HD doesn't really use interlaced formats unless they can be deinterlaced with no impact on the image.
     
  22. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Incidentally, is 720p HDTV 25p or 50p? (30p or 60p for the US)

    If 50p, are the 25 progressive frames doubled (and wouldn't this be a waste of bandwidth) or is the footage actually shot at this high 50p framerate?
     
  23. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    I'd assume the 720p euro format will be 50fps regardless.

    That means if its shot at 25fps each frame will be broadcast twice . If its a film it will be sped up 4% and each frame sent twice or it will just be camera footage shot 50fps. The US system works similarly I believe but at 60fps (24fps , 30fps and 60fps being the actual; frame structure).
     
  24. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    The JVC GY-HD100 is a 720p HD camera. It shoots 24p, 50p and 60p. Will these formats also be stored on tape as segmented frames / even and odd fields?

    Am I correct in saying that there are no video cameras that actually store an entire frame/picture on tape? (I know there's no difference between a segmented frame and a full frame in practice)

    What can one say about the look of 50/60 fps video as opposed to 25/30 fps?
     
  25. Mr.D

    Mr.D
    Well-known Member

    Joined:
    Jul 14, 2000
    Messages:
    11,049
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,126
    Not up on what the hidef cameras actually store . I would have thought the progressive modes store frames not fields.



    50 60 fps would look similar to 50/60i on a progressive display ( ie doing a bob deinterlace) but with much higher detail. Motion wise it would be more characteristic of "video" rather than film.
     
  26. Mandarin Man

    Mandarin Man
    Guest

    Products Owned:
    0
    Products Wanted:
    0
    Ratings:
    +0
    Thanks, Keith. I seem to have run out of questions, now. :)
     

Share This Page

Loading...