I have purchased a XL2, but it has yet to arrive. I plan on shooting a motion picture on it using 16:9 and 24p. I really don't think that it will ever end up on film, more likely dvd. Therefore I was planning on shooting it in 24p 2:3. However, in another topic, I saw this quote "undoable in 2:3:3:2, while it is more or less permanent for 2:3." So I was wondering, should I be shooting in 2:3:3:2 on the off chance I may want to transfer to film? Will I be able to master a dvd or view the source on a regular TV on this setting? If so, why would anyone shoot in 2:3? How is 2:3:3:2 undoable? And if I do shoot in 2:3, will it be a bad idea to try to transfer that to film?

Pardon my ignorance :)

Doug, I'm no expert but think I can answer your questions since I've done lots of research on this, being more or less in the same boat.

2:3:3:2 is reversable because it just creates an extra "fake" frame out of the 24p sequence that can be discarded when editing. As an example, asuming each letter is a field and each group of 2 letters is a frame, the 2:3:3:2 pulldown works like this: AA BB BC CC DD. As you can see, the extra frame created to convert the footage from 24p to 60i is the BC one. Since all its information is present on other frames (BB and CC), it can be discarded by the proper inversion software (a good one is DVFilm Maker) without recompression and therefore loss of quality. That's why this method is "undoable". When you discard the fake BC frame, you're left with the original AA BB CC DD progressive sequence. However, you must understand that this form of pulldown does not look good to your eye when viewed in 60i. Since the extra fields are so close together, it will create stutter and jerky looking images when motion is present. This mode should only be used if you plan to edit in true 24p/23.976p. It is a temporary storing format that is intended to be reversed in post.

On the other hand, the 3:2 pulldown looks much smoother, and is the method used to convert film to tape, but cannot be reversed without loss of data/quality. in a 3:2 pulldown, you'll get something like AA AB BC CC DD... (notice the 3 fields/2 fields pattern) As you can see, there is no way to recreate the B frame without recompression, because it is now dispersed over two distinct frames (AB and BC). This is the smoothest way to make the conversion, but it is not reversable unless you're ready to accept image degradation.

If all you were planing to do is shoot with a "film look" for clients that desire such a look for their wedding for example or other event, the 3:2 pulldown would be fine. However, if there's even the slightest possibility you'll later blow this to film, shoot using the 2:3:3:2 mode. This will allow you to shoot in 24p, transfer your footage on your computer in 60i DV, then discard the extra fake frame of info using a software like DVFilm Maker to get your true 24fps progressive sequence back, edit it in a 24p or 23.976p timeline (check your editing software for compatibility) and then either output it to tape by doing a 3:2 pulldown on your edited work, or create a progressive DVD (or a regular 60i one) or give your 24p footage to a film transfer house for them to do a 1:1 transfer of your work to either 16mm or 35mm. If you're certain all you'll do with your footage is burn it to DVD, then the advantage of a 24p sequence is that you can create true 24fps progressive DVDs for people who have such DVD players and TVs that allow for 24p viewing. If this is your only advantage however, it might not be worth the extra hastle.

Since you can always do a 3:2 pulldown when you're done editing to output to 60i, it's best in my opinion to shoot using the 2:3:3:2 method to keep your options opened. That is of course, if you have the softwares that will allow you to work in a 24p environment or if you're ready to shell a few extra bucks to get some (Adobe Premiere Pro 1.5 and later does both 23.976fps and true 24fps I beleive, don't quote me on that though). If you're on a Mac, you'd probably want to use FCP.

One other advantage to use the 2:3:3:2 mode is that, when you edit in 24p, you have less frames to work on (24fps as opposed to 30fps in an NTSC sequence). Therefore, you can speed up your rendering by 20%. This might not seem like a huge deal but if you don't have the latest and fastest hardware to work on or if, like me, you're a big fan of post-production image correction/transformation and use complicated effects on the vast majority of your footage, you'll be glad you only have 24 frames per second to render instead of 30. Complicated effects can take hours to render.

After that, it's up to you to decide if it's worth it or not. If you're pretty much certain you won't go to film later on, or don't care much for 24p DVDs, and more importantly, if you don't already own the hardware and software that will allow you to work with 24p, I think your best bet would be to stick with the 3:2 pulldown.

Thanks for that post David. That was very helpful to me.

Wow, incredible response. Thanks, that makes things a lot clearer. I am a Mac user and do use FCP, so it looks like I should use 2:3:3:2 since it sounds like its easy enough to master a dvd from that format since it will be edited in 24p anyway. Now I think the challenge is making sure that I configure FCP to convert the stream into true 24p correctly. Do you know if you configure this during capture? Or is it something you do to the video after its captured? Also, when the editing is done, do you have to do anything to convert the stream back to NTSC before compressing it into an MPEG2 stream? Or do you just compress it as is and let DVD Studio Pro know its film material and it'll take care of itself? Of course I'm looking for some mac user expert advise here ;)

Man, I thought I knew something about this stuff... there is so much more to learn. Thanks again!

Doug, I work in a PC environment, so you might want to double check what I say, but to my understanding (again, based on my experience on Windows), you have to aquire the DV stream in regular 60i DV (there's no such thing as 24p DV) and then convert it to 24p using the appropriate software.

If you use DVFilm Maker (check it out here (http://www.dvfilm.com/maker/index.htm)), which costs $145 US, it will then discard the extra frame and give you a true 24p Quick Time sequence (or 23.976fps if you choose that option) that you can edit in FCP. I do not know if FCP can take the extra fake frame out by itself, but you might want to check that out to avoid having to spend $145 on DVFilm Maker for nothing.

As for DVD authoring, I'll let someone with more savvy on that subject answer this one. I do not know if you can convert from your timeline in FCP to 24p MPEG2 directly, but this would be the best case scenario, as you want to recompress as little as possible to avoid image degradation between the different generations.

Just one more question (in addition to those asked above), what is 23.976 fps and why would you use it over 24?

NTSC video runs at exactly 29.97 fps for historical reasons, not 30 (in fact, that's 59.94 interlaced fields per second, but we refer to it as 60i for simplicity). Therefore, if you remove the extra frame in a 2:3:3:2 pulldown, you'll end up with 23.976fps which preserves the audio in sync at 48000Hz.

If you convert a standard NTSC DV signal (29.97p or 59.94i) to true 24fps, the audio will need to be converted to 48048Hz (no noticable difference to the hear, but needed to keep the sync).

This is pretty summed up, but the thing to remember is it's better to use a 23.976fps timeline if you intend to render it back to NTSC. Use true 24fps editing only if you are outputing to film. It would then be easier to do a 1:1 transfer afterwards.

Does Premiere Pro 1.5 allow you to remove the extra frame or would one have to purchase this extra software?

To add to David's superb explanation about 2:3:3:2 vs. 2:3...

It turns out there's another slight advantage to 2:3:3:2. DV compression can actually be slightly more efficient in this mode. DV has two compression methods it can employ, either frame-based or field-based. It chooses which to use based on how much difference it detects between fields. Since progressive frames will show the least amount of difference (because there's no temporal difference) then you'll find the DV compression algorithm will be working in the more efficient frame-based mode for all the frames.

To sum up why you'd want to use which mode, the easiest rule of thumb is:

If you're editing in a 24P timeline, use 2:3:3:2. If you're editing in a 60i timeline, use 2:3.

It really doesn't make a *whole lot* of difference either way... if you're using a 24P-aware editor like Vegas, you can mix and match 2:3:3:2 on the timeline with 2:3 and the program will automatically interpret the footage and remove the pulldown. But 2:3:3:2 is designed for being edited at 24P, can be restored without any loss (whereas 2:3 needs to be uncompressed and have its fields re-combined) and benefits (however slightly) from DV's full-frame compression method. Therefore 2:3:3:2 is better if you plan on editing in a 24P timeline.

David's comment about 29.976 got me thinking. The historical perspective is that the frame rate in NTSC was chosen to be close enough to the B&W frame rate of 30 that B&W sets vertical oscillators would lock to it while also allowing spectral lines of the chrominance signal to fall exactly half way between the spectral lines of the luminance signal so that the chrominance wouldn't cause errors in luminance (chroma phase shifts by 180 degrees from line to line and so the eye doesn't see the error). If you do the math for these conditions the frame rate comes out to 29.97003 and everything I can find on the net relevant to NTSC gives that number. Where does the .006 come from?

It doesn't.

Someone's confusing 29.97 with the telecine'd film frame rate, which is 23.976.

Both numbers are .1% slower than the base number (i.e., 29.97 is .1% slower than 30.00, and 23.976 is .1% slower than 24.00).

Wouldn't 1% slower than 24 be 23.76 and 1% slower than 30 29.7%?

But that's right, NTSC is indeed usually referred as 29.97, not .976. Sorry for the confusion. The 23.976 figure came from DVFilm Maker.

1% slower would be, yes, but NTSC is 0.1% slower.

That's why it's 23.976 (subtract off 0.024) and 29.97 (subtract 0.03).

Sorry Barry, I didn't notice the dot before the 1% in your previous post. I'm sometimes amazed by my own stupidity... ;)

<<<-- Originally posted by Aaron Shaw : Does Premiere Pro 1.5 allow you to remove the extra frame or would one have to purchase this extra software? -->>>

Well Aaron I just upgraded from Premiere 6 to Premiere Pro 1.5 but as far as I know, you still need the extra software (DVFilm Maker) to convert the footage shot in the 2:3:3:2 mode back to 24p.

They added an option in Premiere Pro when creating a new project that allows to select what they call a "Panasonic 24p" timeline to edit at 23.976fps in either widescreen or standard format, and it does a 2:3:3:2 pull-up to render it back to DV, but it doesn't seem to remove the extra frame by itself.

Since I don't have the XL2 yet, I cannot really confirm or deny so I'm not a 100% sure, but so far it looks like this feature is not integrated in the new version. I'll try to find some footage online shot with the 24p 2:3:3:2 mode to edit it in Premiere and see if I'm right, but so far it looks like you would still need DVFilm Maker (Damn you Adobe!). I hope I'm wrong though...

Darn. That would stink :(

Thanks for the info David

Would it really matter in the end? I mean could you still transfer to film using either 2:3 or 2:3:3:2? and if so would one be slightly less quality then the other?

I'm still looking into this so haven't made a personal decision yet. I think 24p 2:3:3:2 wide sounds good but I am also thinking of playing with frame speeds for a particular look, that would mean I couldn't stick with 24p right? and then I would have a difficult time in editing?

Sorry if that sounded confusing

thanks

With two years' experience on this subject on the DVX, let me offer some comment:

If you're intending on doing a film transfer, shoot 2:3:3:2 and edit in a 24P-aware application (such as Sony Vegas). Shooting 2:3 would unnecessarily limit quality.

If you're not going to film, then it all depends on how you're going to edit. If you're editing on a 60i timeline, use 2:3. If you're editing on a 24P timeline, use 2:3:3:2.

Yes you could edit 2:3 on a 24P timeline, and yes it would work, but there would be a minor quality hit that you would take, unnecessarily.

2:3:3:2 can be reconstituted into the original 24P frames without loss. 2:3 requires an uncompress and recombining of fields. But further, the DV compression algorithm has two ways it can encode frames: either field-based, or frame-based. If it detects a lot of interfield motion it'll use the field-based compression, but if it doesn't detect much difference it'll use the more efficient frame-based algorithm. 2:3:3:2 allows it to use frame-based for all the frames. 2:3 involves having to use field-based compression for 3 out of every 5 frames. So 2:3:3:2 can be reconstituted to 24P with no loss and can take advantage of frame-based compression.

Shoot 2:3:3:2 for film transfers. Shoot 2:3 when editing in a 60i timeline, and 2:3:3:2 when editing in a 24P timeline.

How do I find the 2:3:2:3 drop down? I can switch to 16x9 and shoot at 24p but can't find the pull down thing. Am I being silly or is it done in editing? I can't find anything in the manual.

the 2:3:3:2 option is in the XL2 menu. Go to Camera Setup/24p mode select/3:2 or 2:3:3:2

i'm realyy sorry but i can't find it. in camera setup i have:

zebra
skin d set
zoom handle
f speed pset
clear scan
self timer
interval t
int rec time

What about when shooting for digital projection. I'm working on something now that I would like to see digitally projected, and have a space where that could be possible. However, I wouldn't blow this up to film. Does it make a difference then?

Matt

Kevin - It should be the very first option in camera setup.

does a pal xl2 even have it? I'm feeling a bit stupid if so. I'll have another look, thanks. I'll post back if I still have trouble

the first option in menu is time code. next is camera set up which has the things i listed. this is worrying now, i don't get it :/

A PAL XL2 should not have any options for pulldown. The 2:3 and 2:3:3:2 pulldown options are strictly for the NTSC camera. PAL cameras shoot at 25p and record that directly to a PAL video stream, no pulldown necessary.

Thankyou Barry :)

So it is automatically preset for PAL? All I need to worry about is that it is set to 16:9

Barry's right, I didn't realize at the time I was answering to a UK user. Yes PAL models won't have the option. 50i = 25p x2. No weird pull down needed. Every progressive frame comes from exactly 2 interlaced fields.

Well my amateur status clearly shows :D

I've learned quite a bit now from you guys so I hope my first test pieces turn out nice.

I second the digital projection question.

This is a great discussion. Does the fact that you intend to project digitally (an probably not transfer to film) change, what appears to be, the general preference for 3:2:2:3?

Douglas

No. the 2:3:3:2 mode is still the better way to shoot and edit your footage with any kind of progressive screening in mind (digital, film or computer screen).

The only time you might be just as well to use the 3:2 mode is when outputing to TV (and to TV exclusively), but frankly, if you can edit 24p footage and remove the fake frame added by the XL2, I don't even get why you'd use the 3:2 mode. You can always add the 3:2 pulldown in your NLE afterwards if you want to output in a 60i interlaced format. Difference is, you'll also have your intact unaltered 24p footage to work with for non-interlaced applications.

With my XL2 I always use 2:3:3:2 and then remove pulldown in post and work with 24fps. IF I want to go back to 30fps I simply peform 3:2 pulldown and get 30fps again.

Maybe this is an elementary question but here is my situation: I'm going to shoot a commercial for a local church that will play in the local movie theater before a movie begins. I have no idea what they use to project these commercials (maybe it doesn't matter).

My question is - will 3:2 or 2:3:3:2 be preferred in this case? Is someone familiar with how movie theaters project/edit these commercials? The guy I talked to at the theater didn't sound like he knew what was going on. My guess is they throw all of the commercials on a timeline and loop them until the movie starts. Any thoughts?

The choice for 2:3 or 2:3:3:2 is made based on how you're going to edit it, not on how you're going to display it.

If you're editing in a 60i timeline, use 2:3. If you're editing on a 24P timeline, use 2:3:3:2. If you're not going to edit at all, use 2:3.

Barry,

that makes sense - I somehow got the impression from this thread that it did make a difference - I must not have read carefully enough -

I have to shoot tomorrow so thanks for the quick reply.

<<<-- Originally posted by David Lach : Doug, If you're certain all you'll do with your footage is burn it to DVD, then the advantage of a 24p sequence is that you can create true 24fps progressive DVDs for people who have such DVD players and TVs that allow for 24p viewing. If this is your only advantage however, it might not be worth the extra hastle.-->>>

You mean you can create 24p progressive mpg files to put on DVD ? You mean these DVD's can read the progressive 24p and do an internal 3:2 pulldown ?

Or am I missing something ?

Frank

Frank: that is exactly correct. DVD supports 24p through a trick,
any serious MPEG2 encoder and DVD authoring application supports
that features. All commercial (Hollywood) DVD's are encoded in
this format.

Barry: I edited your post, you had written 2:3:2:3 which should
be 2:3:3:2.

DOH! Sorry for the confusion, and thanks for the edit!

<<<-- Originally posted by Rob Lohman : Frank: that is exactly correct. DVD supports 24p through a trick,
any serious MPEG2 encoder and DVD authoring application supports
that features. All commercial (Hollywood) DVD's are encoded in
this format.

Barry: I edited your post, you had written 2:3:2:3 which should
be 2:3:3:2. -->>>

Cool ! One more question Rob. I have a DVD player that can play progressive and an LDP projector. Will it look better if I put 24p mpeg footage on the DVD and have the player to the pullup then doing myself the pullup with software before putting it to DVD mpeg ?

I'm going to try this out RIGHT NOW !! :-)

Thanks again for the info !!

Frank

One more question, if you don't mind.

Do I need to use 24p or 23.976p to make progressive 24p DVD's ?

Thanks !

Frank

You'll want 23.976.

Putting the clips on as 24P yourself, and letting the DVD player do the pulldown for you, is the much better way to go. You'll be encoding 25% less information, which lets you use a lower compression ratio to get better quality on the DVD.

<<<-- Originally posted by Barry Green : You'll want 23.976.

Putting the clips on as 24P yourself, and letting the DVD player do the pulldown for you, is the much better way to go. You'll be encoding 25% less information, which lets you use a lower compression ratio to get better quality on the DVD. -->>>

This is AWSOME info Barry ! Thanks a lot !

Frank

23.976p, but usually the 24p template has droptime which does
that.

Frank: I'd say 24p, but I don't have a progressive player +
screen here, so I really don't know. Should be easy to do some
test yourself to see what works best!

<<<-- Originally posted by Rob Lohman : 23.976p, but usually the 24p template has droptime which does
that.

Frank: I'd say 24p, but I don't have a progressive player +
screen here, so I really don't know. Should be easy to do some
test yourself to see what works best! -->>>

Sadly ! I'm using Vegas 4.0 and it somehow can't save a 24p mainconcept mpeg2 file. I stutters !

Also DVD Architect 1.0 can't handle 24p mpeg2 files. it always wants to rsample it to 30i with pulldown.

Once I get my XL2 I'm going to get The latest Vegas + DVD package 5.0. Hope all the problems will be fixed then. I'll know more once I visit DVExpo this week ! :-)

I DO have a progressive DVD and a good projector. I have the hardware, but not the software yet ! :-)

Thanks for all the help though. I'm learning LOTS !

Frank

Vegas5+DVD can totally handle a pure 24P DVD production. On the DVX DVD I encoded some segments as 24P, and some as 60i (specifically the segments that showed 60i vs. 24P) and it did just great with all of it. Once you upgrade to Vegas 5+DVD you should be able to make 24P DVD's all day long.

<<<-- Originally posted by Barry Green : Vegas5+DVD can totally handle a pure 24P DVD production. On the DVX DVD I encoded some segments as 24P, and some as 60i (specifically the segments that showed 60i vs. 24P) and it did just great with all of it. Once you upgrade to Vegas 5+DVD you should be able to make 24P DVD's all day long. -->>>

Woohoo !

Can't wait till February ... that's when I get my bonus ... and the XL2 ! :-)

Frank