Pull down cadence if that's the right terminology?

L

lukehumphrey

Active member
Can someone demystify what is meant by the concept of pull-down cadences or point me to a good web reference? I was playing with the settings on my TV and the Image menu had an option to enable 3:2 pull-down. Also I know the settings on my DVX100a have different pull-downs (I guess) for F5-6. I'd like to understand the concept of pulldowns abstractly and how they relate to usage of my dvx100a. But more emphasis on pull-downs generally/conceptually. E.g. why are they necessary? why doesn't the hardware simply work like film where it captures progressive frames and flips through those frames quickly to simulate motion? why the complexity with pull-downs and different shooting modes!? stuff like that :).
 
The necessity of pull down stems out of the desire to extract the fundamental (progressive) frames from within an interlaced stream (like NTSC). There are two aspects that should be understood.

First, all NTSC signals are interlaced and each frame is made up of two fields. One of these fields is the odd scan lines, the other is the even. A progressive source is broken into to fields, one which is comprised of 1/2 of the frame (every other line) the other is comprised of the oppsite lines. This would break a 24 frame per second film into 48 fields per second.

Second, all NTSC signals are 60 fields per second (30 frames per second). Since the 48 fields that would be created in the example show above are not compatable with the 60 field nature of NTSC, something must be done to make it compatible. The method is to repeat every other field. This yields 24 fields (odd scan lines) + 24 fields (even scan lines) + 12 repeated fields. The sum of 24+24+12 is 60 fields which now makes the 24 frame per second material fully compliant with the requirement of NTSC. The pattern of these fields is referred to as the cadence.

Knowing which fields come from which original frame, it is a fairly simple matter to reassemble them into the original 24 frames.
 
This gives a graphical view:
http://www.adamwilt.com/24p/index.html#24pRecording

Pulldown (as in film being pulldown) is used to convert 24 fps to 60i video. Fields are duplicated and combined to maintain the smoothest motion possible with the fewer original frames.
This is traditionally done when film is telecined. 2:3 refers to the cycle of frames/fields.
Converting the resulting 60i back to the original form of 24fps is removing pulldown (ror reverse telecine)
 
Ok, thanks. This is helping. It's begninning to make sense to me. NTSC is a standard that will always require being played at 30 frames/second, 2 interlaced fields per frame. Progressive scanning tries to work with those format requirements to get a non-interlaced look. Progressive video is still technically interlaced, but it's done in way to make it look progressive, or like solid autonomous frames flipping by quickly. So within the context of the information you've given me, what do the following mean:

1. My 16x9 HDTV: Enable 3:2 pulldown
2. In Avid 24p and Film Settings: Standard pulldown 2:3:2:3
3. In Avid 24p and Film Settings: Standard pulldown 2:3:3:2

And if 24p is still just interlaced odd/even fields but some repeated fields thrown in to "virtually" slow down the frame rate, what is 30P? Wouldn't that be the same as 60i, 30fps with 2 odd/even fields per frame?
 
Not to take away from my last post (breaking those down will really help me), but reading the adamwit link above, it looks like every other 24p has a different number of fields associated with it. So the first frame has 2 fields, the second 3 fields. how does having these "short" and "long" frames not make the motion look off? isn't this sort of tantamount to a film projector holding on every other frame slightly longer?
 
The two cadences used are either the repeating pattern (24p - normal) 2:3:2:3...

or

2:3:3:2:3... (24pa - advanced)

Of these two, 24p normal looks best when viewed on an interlaced playback system.

The advanced method is intended for applications that are progressive in nature (NLEs that are advanced capable, film out, progressive DVDs, etc).

30p is unique in that it doesn't have extra frames added, rather it is just progressive material delivered within an interlaced stream. It looks fine when viewed on an interlaced monitor and is easily handled by NLEs.

The underlying reason for the existence of these techniques is to allow progressive footage to be carried within an interlaced system.

The option for your TV is to enable pull down detection (which has to do with its internal de-interlacer) and would normally be selected to yield the best results.
 
There's some confusion here. 24p and 24pa both capture true frames. There's no interlacing involved. Now in order to go out to MiniDV tape running at 60i 9or even video to your NTSC monitor) they have to add a conversion.

24p adds a telecine style pulldown. 2:3:2:3 This means that the progressive image is broken into fields, some fields are duplicated and the final output is a smooth 60i version of something captured at 24fps pure frames. this is the exact process film goes through when transferred (telecine) to video. In this mode you can capture, edit and display as normal video but it will have the same inherent film motion.
If you need to convert this back to 24fps you remove the pulldown. Unfortunately since some of the frames have been split into fields and combined the software has to de-interlace and recombine the frames and then recompress these manipulated frames. This takes some time and can cause some slight quality loss.

24pa is converted in the DVX to 2:3:3:2 pulldown format. Similar in process to above but it doesn't provide the smoothest motion at 60i. It's whole point is to allow the editing software to quickly and easily drop these extra frames to go back to 24 fps. This provides the best quality.

24 fps final results requires pulldown removal of either type and is primarily for film output or 24fps DVD. 24fps DVD (like current films) requires the DVD player to add pulldown 2:3:2:3 when outputting to a normal interlaced TV.
 
Scottdvx100 said:
24 fps final results requires pulldown removal of either type and is primarily for film output or 24fps DVD. 24fps DVD (like current films) requires the DVD player to add pulldown 2:3:2:3 when outputting to a normal interlaced TV.
Click to expand...

A DVD which is encoded with an MPEG stream derived from progressive sources (film or progressive video) also has the advantage of not needing to contain the duplicate fields (rather it use a repeat flag in its place) in the GOP (group of pictures) and therefore allows for more material and / or higher average data rates. These types of discs result in better looking footage when played back on either an interlaced or progressive playback system.
 
Thanks everyone for the great info. Scott, I wanted to follow up with you on "24fps DVD (like current films) requires the DVD player to add pulldown 2:3:2:3 when outputting to a normal interlaced TV."

So by 24fps DVD you're talking about progressive scan DVD players that seem to be fairly prevelant. Now the TV part - do tvs exist that are not the normal interlaced, but can play progressive video? I hope that made sense, still trying to wrap my head around the concepts and relate them to my personal practical application :).

Also, can I assume that if my project's final output is DVD to be played on a progressive DVD player through an interlaced tv, using the 24p advanced is the best way to record footage with my dvx100a?
 
A DVD which is based upon progressive scan source material does not require a progressive scan DVD player to realize the advantages of this type of MPEG encoding. Regardless of the nature of the MPEG encoding, pull down is added for interlaced output and deinterlacing is used for progressive output. It is simply that the progressive scan source material that is encoded as 23.976 frame rates is more efficient in terms of compression (conversion into an MPEG stream) than progressive scan material that uses the 2:3:2:3 interlaced cadence, since the 23.976 frame rate material doesn't add any extra fields to the GOP.
 
I understand the flagging for the dup fields to save space, I think that's great. Currently I use sorensen squeeze to mpeg encode my 23.97 footage, but I encode to NTSC 29.97 - can and should I instead encode to 23.97 to get a truly "progressive DVD"? If I do this is it best then to record my source footage using 24pA because my progressive DVD will get better quality when the DVD player interlaces the progressive source DVD for output on my tv?
 
I don't use the Sorensen tool, so I don't know if it handles progressive footage properly, but can offer a test to determine if it does. Just build a short .m2v stream that has been setup for 29.97 NTSC MPEG2 DVD application. Now, don't change the compression settings, just the GOP for 23.976. Compare the two file sizes. If handled properly, the second .m2v stream will be about 20% smaller in size.

As for the second part of your post, the DVD player doesn't interlace the stream; rather it adds pull up to make the stream NTSC compliant. This is accomplished prior to the video encoder (video DAC section) within the DVD players MPEG decoder.
 
Thanks I'll try that.

Regarding the second part of my question, thanks for clarifying. Am I correct in assuming, however, shooting 24p Advanced will allow better end picture quality when viewing progressive DVDs on an NTSC tv, rather than shooting 24p Standard?
 
Very, very, very slightly better, maybe. But probably not enough that anyone would ever notice.

Basically if you're editing in a 24p timeline, shoot 24pa always. If you're editing in a 60i timeline, or don't know how you'll be editing, or you're not the one doing the editing, that's when you would choose 24p Normal.
 
Doesn't matter what the final target is, it matters how you intend to edit it. If you're editing in a 24p timeline, 24pa is better. For other purposes, 24p is better. It's not the final output that matters, it's how you intend to edit.
 
Barry_Green said:
Very, very, very slightly better, maybe. But probably not enough that anyone would ever notice.
Click to expand...

Hey Barry,

I find this comment a bit curious, but perhaps I am misreading your response.

Are you saying that you don't find that using a shorter GOP with progressive material makes much qualitative difference on the resultant MPEG stream? I have been doing authoring since early 1998 on a number of different platforms, but most on SGIs running Scenarist (from the pre Sonic days) and have to say, that with difficult material, I find the advantage noticeable. With the same peak data rate limits you can end up with a stream which is closer to the source material or you can choose to lower the average (holding the same quality) and have less impact on your bit budget.

I would be very interested to learn your comment on this as this differers from my experiences.
 
I have no idea what you're talking about in relation to the statement I made, so obviously we're not talking about the same thing...

The question was: should I shoot 24PN or 24PA in order to author a 24P DVD, and will 24PA give me a better-quality final product. My answer was: only the tiniest bit better, because 24PA does benefit slightly from frame-based DV compression so the source will be just a little tiny bit cleaner. After that you're still going to edit the same way and author the same way, so the rest of the chain's all the same.

If your point was "don't you think a 24p dvd looks better than a 24p-within-60i DVD", the answer is oh heck yes. But I was assuming he was going to edit and master to a 24p DVD anyway. So the only difference would be in the quality of the source footage, and the 24PA progressive-recording system would give a small improvement in individual frame quality vs. a 24PN source.
 
Thanks Barry,

Yeah, I totally misread your reply and I absolutely agree.

Thanks for the clarification.
 
I think the final target and editing go hand in hand. If you don't require a pure 24fps output at the end of the project there's not a lot of reason to edit in a 24fps timeline or to choose 24pa mode.
The final target and look will determine what your inital shooting format is. (60i, 30p, 24p/24pa)
If you plan to only go out to standard video but want the 24p look then there's not much reason to go through the removal pullodwn , re-compression, edit 24 and then to add pulldown.
 
You must log in or register to reply here.
Back
Top