So you would use 3:2 Pulldown With? at DVinfo.net

Roman Stikel · October 25th, 2010, 06:18 AM

Ok, I thought I would simply start another thread. Then, you would you 3;2 pulldown with a particular camera or what?
Thanks Roman

Dan Keaton · October 25th, 2010, 07:22 AM

Dear Roman,

If you have a camera that supports 24p or 23.976p,
and you setup your camera to record in that format,
then you would enable the Video|Remove 3:2 Pulldown
so that the nanoFlash is instructed to remove the duplicated frames.

This option saves space on your CompactFlash card as the duplicated frames are not recorded.

And this can make editing of the files easier as the duplicated frames do not have to be removed in post.

Mark Job · October 25th, 2010, 07:51 AM

Hi Dan:
If the camera already does 24p, then it is not necessary to use 3:2 pull down removal. because there is no redundant extra frames to remove. Is this a case where you are referring to 24p as not actually 24p, but 23.976p ? - If that's the case, then your statement is even more confusing, because no camera actually outputs 23.976p to my knowledge. However, if your camera is an XL H1, then it outputs 24F, which is actually 23.976p *Only After* 3:2 pull down removal has been activated on the Nano or XDR, which gives you 23.976p.

Aaron Newsome · October 25th, 2010, 01:38 PM

I use it with any camera that sends a 24p signal inside a 60i signal. 3:2 pulldown removes the duplicate frames from 60i, giving you the 24p back.

Dan Keaton · October 25th, 2010, 02:51 PM

Dear Roman,

I can confirm what Arron said.

And, if one has a camera that does not support 24p (or 23.976p),
then one should not enable Video|Remove 3:2 Pulldown.

Billy Steinberg · October 25th, 2010, 04:17 PM

Roman,

While you are getting correct technical answers to your question, I wanted to comment on "why/when would one want to use 24 frame rather than 30 frame?" in a (slightly) less technical way. (I am speaking in even frame numbers throughout this message, as decimal frame rates aren't relevant to this discussion, but note that we're probably really talking about 23.978 and 29.97).

Film is usually shot at 24 frames per second.
US Video is usually shot at 30 frames per second.

When a movie was transferred to video, there had to be some way to get the film shot at 24fps to display at 30fps without running off speed. If US television had a frame rate of 48fps, it would be an easy conversion, just show each film frame twice before going on to the next film frame. But 30 divided by 24 doesn't work out evenly like 48 divided by 24. So a complex formula was devised to convert 24fps to 30fps. One of the trade-offs to the method used results in what's called "motion judder". Smooth motion is no longer smooth. Every few frames in the video there is an extra frame added. And it's not done evenly; three frames, then a duplicate, two frames, then a duplicate, etc. That's where the 3:2 pulldown expression comes from. (And there are other pulldown cadences that can be used besides 3:2, and then there's interlaced video, with it's two fields per frame, which adds complexity too).

All this was about getting 24fps film to play at the proper speed when shown on 30fps television. And this has been going on since the beginning of television; it's nothing new. And with one marvelous exception, none of this has anything in particular to do with the nano*. I am also leaving out PAL TV frame rates; they don't have nearly the problems we do here in the states because their video is 25fps and the 24/25fps conversion is much simpler.

Something that everyone "in the business" soon noticed was that people reacted differently when watching content that was converted to 30fps from 24fps. Without being able to define why, most people felt that watching true 30fps had a "live tv" feeling, and watching 30fps video made from 24fps source didn't elicit that feeling at all. Some felt that native 30fps felt "cheap" and converted 24fps felt "expensive". Advertisers always shot their tv commercials in film until recently, and years ago they experimented in shooting their film in both 24fps and 30fps. Same commercial, same film cameras, same lighting, same actors, etc, just shot at both frame rates. Then they transferred them to video, and tested both versions, and found their test audiences reacted differently to the two versions. So depending on what the advertiser wanted, the commercials were shot at the appropriate frame rate from then on.

So why has this become news all of a sudden? Two reasons. Video cameras now provide 24fps as a standard frame rate, which they never did in the past, and distribution of completed videos is no longer limited to viewing on a tv set with a CRT display, which had a single frame rate available, 30fps. A significant portion of video is now watched on computers, which can display video at multiple video rates, and most tv sets made in the past 5 years or so don't use a CRT and can display (and convert, if necessary) any frame rate.

So the first (and most significant) question to ask yourself, is what is the final edited video going to be used for. How is it going to be viewed and distributed? If it's primarily going to be seen on TV, then maybe you want to acquire in 30 frame; if it's going to be seen mostly from a blu-ray disc, in hi-def, on modern TV, then maybe you want to shoot in 24 frame.

*As far as the marvelous feature the nano can offer you, many current cameras can acquire (shoot) in 24fps, but since most tv sets can't display 24fps, only 30fps, the camera manufacturer builds the 24fps to 30fps converter into the camera, and the signal recorded in the camera and the nano is 30fps, with all the duplicate frames already added. Recording duplicate frames is wasteful (20%), and editing 24fps in a 30fps wrapper is much more difficult. You can't just cut to the frame anymore, you have to cut preserving the pulldown cadence. So the nano has a mode that will take the pre-converted video from the camera and turn it back into raw 24fps before recording it. (Again note that when I say raw 24fps here, I don't mean as opposed to 23.978, I mean as opposed to 24fps embedded in 30fps video).

And since very few monitors will display 24fps, CD has built an output converter into the nano as well, so you can play your raw 24fps video on a normal monitor. You feed the nano 24fps embedded in 30fps wrapper, it extracts back to raw 24fps video, records it, and then it re-embeds it back into a 30fps wrapper so you can monitor it easily.

As a side note, many of the live concerts I shoot are done in "30p" as opposed to "24p". This gives you most of the desired parts of shooting 24p with none of the disadvantages. A different look and feel from "standard" 1080i without the uneven motion judder of 24p. But that's a whole other discussion...

Hope this helps, rather than confuses.

Billy

Brian Doherty · October 25th, 2010, 05:06 PM

As an interested and daily reader of this forum for a couple of years now, I value Billy's contributions as they strike me as being both balanced and helpful.

Even though I have been "in the business" for over 40 years, I find insights such as those provided by Billy and others, offered up in the spirit in which they are, very useful in furthering my understanding of the complexities of modern day film/television/video operations.

To Billy and others who offer advice in such a positive and constructive way, please keep up the good work.

Brian Doherty

Dan Keaton · October 25th, 2010, 07:04 PM

Dear Billy,

We greatly appreciate you sharing your knowlege.

What a wonderful post!

Daniel Symmes · October 25th, 2010, 07:40 PM

The decision really comes down to:

If the film is to be shown theatrically, you shoot 24p (TRUE 24).

If you are not theatrically bound, you can choose 23.976 to produce the "film look" and be capable of TV viewing (which 24p isn't).

If you want the best in image clarity of motion, AND TV broadcast, select 29.97.

There is also 30p which is less common in usage (broadcast requires 29.97 or 23.976 with 2-3 pulldown).

There are conversion methods between the various frame rates, but often have negative side affects.

I just shot a special video for DreamWorks. They decided on 29.97 only to find (the next day) it was destined for theatrical projection!! To make it worse, we shot 3D, and most methods for 29.97>24 don't produce identical L&R views.

Billy Steinberg · October 25th, 2010, 08:22 PM

Daniel: When I mentioned "30p", note that I deliberately was NOT getting into the 1/1.001 area of things. The stuff I shoot in "30p" is really 1080 29.97 psf. It is definitely supported everywhere that 1080 29.97 i is supported. (In fact, there is no way to tell the difference by looking at the program stream, though you can tell by watching the picture. Note that one's recorder should be told whether the format is interlaced or progressive segmented frame so its codec can work most efficiently. Even the Sony SR decks want to be told).

Many of my clients who used to specify 23.97psf for its effect now specify 29.97psf. I just shot a series of ballet performances for PBS (a 1080i network) in which one of the performances was in a cartoon-like set, and for that performance I set the cameras to acquire in 29.97 progressive. They wanted that performance to look less "real/live" but didn't want to add motion judder to ballet.

Billy

Daniel Symmes · October 25th, 2010, 08:47 PM

Billy -

Thus my comment that 30p is uncommon. Like 24p, there are uses.

Shooting for "video" viewing is simply 23.976 or 29.97.

A major video post house in LA used to mess me up with the 23.98 confusion. They actually set their equipment to that rate and kept messing up my audio. They didn't get the distinction that 23.976 is a real number (well...), with 23.98 being short-speak.

Interlaced formats are another bag. I do a lot of 3D, and because of post alignment needs, interlaced is a drag.

PsF is a another drag to many. Invisible issue until you find a monitor that needs it (or doesn't).

But, we have to be ready to give our clients what they want, even if it isn't ideal.

Dan Keaton · October 26th, 2010, 08:10 AM

Quote:

Originally Posted by Mark Job

Hi Dan:

If the camera already does 24p, then it is not necessary to use 3:2 pull down removal. because there is no redundant extra frames to remove.

Is this a case where you are referring to 24p as not actually 24p, but 23.976p ? -

If that's the case, then your statement is even more confusing, because no camera actually outputs 23.976p to my knowledge.

However, if your camera is an XL H1, then it outputs 24F, which is actually 23.976p

*Only After* 3:2 pull down removal has been activated on the Nano or XDR, which gives you 23.976p.

Dear Mark,

There are cameras that produce 23.976p without adding 3:2 pulldown. With these cameras, one does not need to enable our Video|Remove 3:2 Pulldown option (obviously).

There may be cameras that produce true 24p without adding 3:2 pulldown.

And there are cameras that produce 23.976 in a 1080i59.94 stream by adding pulldown frames.

And there may be cameras that produce true 24p in a 1080i60 stream by adding pulldown frames.

Page 56 of the English version of the original Canon XL H1 manual has information on the frame rates available for the camera.

For this 1080 only camera, they support 60i, 30F and 24F.

(Canon uses the term F instead of P for progressive since the sensor in the XL H1 is an interlaced sensor.
But, as has been proven many times on this wonderful forum, the F modes are really progressive.)

Their terminology they use is as follows:

Their 60i is actually 1080i59.94, (which is actually 29.97 frames with 59.94 fields)
Their 30F is actualy 1080p29.97,
Their 24F is actually 1080p23.976

The above information was just confirmed via Canon's technical support and applies to the US version of the camera.

When the XL H1 is upgraded to include PAL support, the PAL frame rates are different, as 25 Frames per second in PAL areas of the world is true and not a fractional number. And 50i is true also.

Mark Job · October 26th, 2010, 06:55 PM

Quote: