Well, folks, I do believe I'm about to make the leap into theatrical production. Does anyone have experience (or links to reliable opinions) on which is preferable for eventual film out: 1080PA or 720 PN? Normally I'd go for the 1080, but I'm not sure whether the pulldown is more objectionable than the lower resolution.

FCP can edit 1080 24p, just use Cinema Tools to get rid of the pulldown.
As far as the actual difference? I say do a test and shoot the same things in both modes.

Many have noted that 1080 gives higher resolution than 720. But you can't over/under crank in 1080.

If it were me I'd go for 1080p 24fps if it's a straight shoot and you want the highest resolution, but I'd go for 720PN if varying speeds are important for your target style.

Thanks... yeah, I don't expect to need VFR on this project. I suppose I'll end up having to shoot some tests, but at the cost of 35mm film-out (not to mention the problem of finding a screening room to project it here), I was hoping to find someone who'd already done it. I've noticed some very apparent pulldown frame artifacts in some video-to-film projects (notably Murderball), but I'm not sure if it's regular pulldown or if I'm seeing some other motion effect in post on top of the pulldown.

Unless I'm wrong, there's no way to shoot 1080/24 (on the HVX) without there being pulldown on certain frames of the final film-out. I think 720/24PN is the only way to get a true clean frame for each frame of film. (If I'm reading the pulldown diagrams wrong, by all means tell me!)

I'm also reading that some inverse-telecine processes soften the image to help mask the interleaved pulldown frames.... and that the end result is resolution no better than 720/24PN.

If anyone has access to real-world test results on this (not necessarily specific to the HVX... this applies to all 1080/24 over 60 digital cinematography), I'd sure appreciate it. This may be a call for the CML...

Unless I'm wrong, there's no way to shoot 1080/24 (on the HVX) without there being pulldown on certain frames of the final film-out.
Why would you think that? That's completely not accurate. 1080/24pA can be edited in FCP or EDIUS as native 24p with no pulldown artifacts of any type, just straight raw 1080/24p. And you can extract the pulldown out of 1080/24pA footage in Raylight to end up with pure straight 24p.

I think 720/24PN is the only way to get a true clean frame for each frame of film. (If I'm reading the pulldown diagrams wrong, by all means tell me!)
Okay, I'm telling you. You can get exactly as clean frames from 1080, you just have to use a program that knows how to extract the pulldown.

barry, question then...
other than variable frame rate, and if the color renditions of both 720p and 1080p are the same for this camera, why would anyone opt for 720p over 1080p, especially for a filmout...

It might work better for workflow as you can get twice as much time on a P2 card or whatever format you are storing on with 720 rather than 1080.
This can be a significant factor.

Why would you think that? That's completely not accurate. 1080/24pA can be edited in FCP or EDIUS as native 24p with no pulldown artifacts of any type, just straight raw 1080/24p. And you can extract the pulldown out of 1080/24pA footage in Raylight to end up with pure straight 24p.

Well, 'cuz I'm a dope, I guess ;-) And your book hasn't arrived yet!

What I've seen certainly seems to point to the frames in 24-over-60 as being interpolated frames created by blending various pairs of fields from the 60i recording.

There's only two ways to create a frame from 60i fields...either you have a raw frame with the interlaced-field jaggies on any areas of movement, or you have a frame that's been artificially interpolated by smoothing out the areas of difference between the odd and even fields. I suppose some of the really high-end software does that interpolation with great results, but I was concerned about softening or edge artifacts.

Since the camera itself is recording 60i the whole time (in 1080), I assume the creation of 24 progressive frames from those 60 fields happens in post, and the results probably vary depending on what system you're using. I'm primarily an Avid guy, but the Adrenaline I work on doesn't have the HD expansion board yet, so I hadn't tested it. I own a copy of FCP, but I'm really just at the flailing-around stage of learning it.

When you say "extract the pulldown out of 1080/24pA footage in Raylight to end up with pure straight 24p" I assume you're still talking about interpolation. That's really ~not~ the same thing as recording true 24p. And I always assumed the 720/24PN setting was what I'd consider "true" 24p--individual progressive frames created in-camera straight off progressive CCDS without jumping through mathematical hoops to blend two interlaced fields separated in time.

Am I makin' any kind of sense here?

An additional point to consider in the 720 vx 1080 decision is are you going to do any CGI work in post with your footage? Results from side-by-side tests showed that a cleaner matte can be pulled in the 720p mode. So, even though the filmout may have a bit better resolution, if there are greenscreen shots, the artifacts will be more evident tif shot in 1080.

As far as the pulldown cadence not allowing for a "clean" 24p output for the filmout, I think you are not quite accurate in how the progressive frames are stored on the P2 format. I believe that in the 1080p24 format, a progressive frame is pulled off of the CCDs with the same clocking, and then it is split into two fields and stored in concurrent frames interlaced. Barry has already outlined how that process is reveresed in post and true progressive frames are reassembled.

If interlace artifacts are in a filmout version from progressive footage, the likely culprit is both lazy reverse telecine, and not catching a bad pulldown cadence setting of a clip.

Cheers

What I've seen certainly seems to point to the frames in 24-over-60 as being interpolated frames created by blending various pairs of fields from the 60i recording.
No, that's backwards. It's interlaced frames created from splitting apart progressive frames. Exactly the same way as film gets telecine'd to video.

There's only two ways to create a frame from 60i fields...either you have a raw frame with the interlaced-field jaggies on any areas of movement, or you have a frame that's been artificially interpolated by smoothing out the areas of difference between the odd and even fields.
On other cameras, maybe. But not the HVX. The HVX doesn't start with interlaced frames that it tries to blend into a simulated progressive frame; that's what Canon and Sony do. The HVX starts with a progressive frame, and only a progressive frame. It records those progressive frames within an interlaced data stream, because 1080 DVCPRO-HD is only interlaced.

Look at 1080/24pA footage, you'll see there are four straight progressive frames, followed by one "split frame". That "split frame" is formed by using half of the frame before it (the even lines, presumably) and half of the frame following it (the odd lines, presumably, or maybe it's the other way around). Point is, that "split frame" is only there to "pad out" the sequence, to get 24 frames padded out to 30. Drop that split frame and what are you left with? 24 original pure progressive-scan frames.

When you say "extract the pulldown out of 1080/24pA footage in Raylight to end up with pure straight 24p" I assume you're still talking about interpolation.
Nope. I'm talking about reversing the "telecine" process to restore the frames to their original 24p nature. In 24pA you do that by just ignoring the artificially-created "split frame". In 24P mode (3:2 pulldown) you have to divide the frames back up into fields and re-sort them to assign the right fields together, and once you do that you can treat them as progressive frames (because they're NOT fields, they are halves of a progressive frame).

That's really ~not~ the same thing as recording true 24p.
The way the DVX, XL2, and HVX do it, yes it is.

And I always assumed the 720/24PN setting was what I'd consider "true" 24p--individual progressive frames created in-camera straight off progressive CCDS without jumping through mathematical hoops to blend two interlaced fields separated in time.
It is. And 1080/24p is exactly the same thing, and so is 1080/24pA. It just gets recorded in a 2:3 pulldown or 2:3:3:2 pulldown so it "fits" in the interlaced data stream, but it's progressive footage and can be restored back to its original pristine progressive nature very easily.

An additional point to consider in the 720 vx 1080 decision is are you going to do any CGI work in post with your footage? Results from side-by-side tests showed that a cleaner matte can be pulled in the 720p mode.
I haven't seen those tests, but I can say that something's wrong, that isn't true nor can it possibly be true. Someone messed up somewhere.

1080 mode contains the identical same image, the identical same picture, the identical same color and the identical same latitude, the identical EVERYTHING, as 720 mode does. It just has more of it. Both modes start out internally at 1920x1080, and go through the exact same processing. Everything is identical between them right up until the instant that it gets recorded. At that point the 1080 mode gets recorded at 1280x1080 (or 1440x1080 in the 50Hz version), and those resolutions are high enough to retain almost every bit of detail the HVX can generate. Whereas in 720p mode the image gets scaled down to 960x720 before recording. That results in some detail loss.

So, even though the filmout may have a bit better resolution, if there are greenscreen shots, the artifacts will be more evident tif shot in 1080.
Disagree 100%. Everything about 1080 is just better. Any artifacts that show up will be smaller in 1080 mode because the image is spread across a frame that's twice as large.

If interlace artifacts are in a filmout version from progressive footage, the likely culprit is both lazy reverse telecine, and not catching a bad pulldown cadence setting of a clip.
Exactly correct!

other than variable frame rate, and if the color renditions of both 720p and 1080p are the same for this camera, why would anyone opt for 720p over 1080p, especially for a filmout...
Leonard answered this -- the only reason to opt for 720 is if you need more recording time on the cards. You trade a small (very, very small) reduction in picture sharpness in exchange for getting 2.5 times as much recording time. Lots of people seem to think the tradeoff is well worth it.

But if you're looking to extract the absolute maximum from the camera, 1080p does that.

Well, folks, I do believe I'm about to make the leap into theatrical production. Does anyone have experience (or links to reliable opinions) on which is preferable for eventual film out: 1080PA or 720 PN? Normally I'd go for the 1080, but I'm not sure whether the pulldown is more objectionable than the lower resolution.

Perhaps you should try starting out on a less technically challenging project. There's no better teacher than experience.

Perhaps you should try starting out on a less technically challenging project. There's no better teacher than experience.

Too late! ;-)
I'll run tests, but I've been doing "less challenging projects" for 20 years. I've raised the money; the project is a 'go.'

There's really nothing challenging about it... I'm just trying to figure out the pro/cons of the two formats.

No, that's backwards. It's interlaced frames created from splitting apart progressive frames. Exactly the same way as film gets telecine'd to video.

--snip--
On other cameras, maybe. But not the HVX. The HVX doesn't start with interlaced frames that it tries to blend into a simulated progressive frame; that's what Canon and Sony do. The HVX starts with a progressive frame, and only a progressive frame. It records those progressive frames within an interlaced data stream, because 1080 DVCPRO-HD is only interlaced.



Ahhhhhhhhhhhhhhh
This was totally NOT clear from the material I'd read.

Panasonic's own chart of recording times/formats describes 1080/24 as being over 60i and 720/24 as being over 60p. Nowhere did I find a description of exactly what was going on in terms of how the signal was processed, though.

What I implied--incorrectly, obviously--was that in 1080 mode the HVX worked like the Sony and Canon, and in 720 it was a true progressive signal.

Thanks for clearing THAT up!

...Everything about 1080 is just better...The only exceptions to this are the additional storage space requirement and processing load, as Leonard mentioned. I think it impacts these three areas:

1) Half as much video fits on each P2 card. No big deal, if you're doing short takes, have enough cards, and have a good workflow for moving the bytes around.

2) You need twice as much hard disc space. Fortunately HDDs are cheap, and

3) The CPU has more work to do. As long as you have a strong enough system to preview things well, and allow enough time for your renders, it's likely not a problem.

If quality is more important than a bit of time and money, and you don't need variable frame rates, I'd shoot 1080.

And Barry's absolutely right about the true 24p thing. Think of it as a 24p product shipped in a 60i box. The extra fields that get thrown away are like Styrofoam packing materials - only they're easier on the environment. :)

Leonard answered this -- the only reason to opt for 720 is if you need more recording time on the cards. You trade a small (very, very small) reduction in picture sharpness in exchange for getting 2.5 times as much recording time. Lots of people seem to think the tradeoff is well worth it.


If I were going to HDTV, I'd agree. But for 35mm film-out, I think it's worth testing. Even minor differences in resolution aren't so minor when the image is that big. The HVX will only be the B-camera, but it'll be on-screen a lot of the time.

Think of it as a 24p product shipped in a 60i box. The extra fields that get thrown away are like Styrofoam packing materials - only they're easier on the environment. :)

Makes perfect sense now. It had just never been explained to me. Thanks!

[QUOTE=Barry Green]I haven't seen those tests, but I can say that something's wrong, that isn't true nor can it possibly be true. Someone messed up somewhere.

I'll see if I can get some examples of the test. It was run by some of the folks at PixelCorp who tested the HVX in 720PN, HVX in 1080P, Varicam, F900, and F950. The sole purpose of the test was to evaluate performace of the footage for CGI work. A quote from one of the sr.testers "...the edges of the 1080p footage from the HVX was -softer- and harder to get a crisp matte, given the option use 720PN..."

Sure there are lots of variables with tests, and I did not personally perform it. But I did talk with one of the testers who is a friend of mine and who I trust. I'll see if I can get more details of the tests to post so you can draw your own conclusions.

Cheers

My 2 cents here.

Test what you will be shooting in the format you want to shoot- We shot our feature at 720 - we performed a film out test with that and it stood up very well to a film out. So the added storage and processing and reduced p2 time of 1080 werent justified - we were thrilled with the quality of 720.

Make sure you need 1080 if you use it - if you dont need it you will be swapping cards more then you need to and peforming backups and backups more then you need to - the on set workflow is not trivial.

see my report on the filmout here
http://www.dvxuser.com/V6/showthread.php?t=51195&highlight=filmout

and you can see trailers from the film here (uncolor corrected) - http://www.americanstandardthemovie.com/media.html

...A quote from one of the sr.testers "...the edges of the 1080p footage from the HVX was -softer- and harder to get a crisp matte, given the option use 720PN..."I think I can explain this...

If you're looking at a given screen size, 1080 gives you more resolution PER INCH than 720. It doesn't give you the full 50% resolution improvement that is theoretically possible, due to the limited resolution of the imager. People here have estimated that you get about 20% more resolution with 1080 than 720. This could be important for film out. The movie-goer will potentially see a 20% sharper image.

Now zoom in to the pixel level. 1080 is 1.5 times more dense in one dimension than 720, yet it's only 1.2 times sharper. That means that the 1080 version is only 80% as sharp as 720 PER PIXEL (1.2 / 1.5 = 0.8).

Chromakeyers don't care how big the screen is. They work on a local group of pixels. While the audience sees 1080 as sharper across the screen, the chromakeyer sees 1080 as softer on a pixel-by-pixel basis.

All things being equal, the chromakeyer would give a thinner green outline (per inch) on 1080 than 720, assiming both are projected on the same size screen. However, if the soft transition confuses your chromakeyer in 1080, but the sharper 720 transitions are interpreted well, 720 might give the better chromakey than 1080.

That's my theory anyway.

BTW, I rent an HVX tomorrow. It's time for me to dump the theory and get practical for a change!

My 2 cents here.

Test what you will be shooting in the format you want to shoot- We shot our feature at 720 - we performed a film out test with that and it stood up very well to a film out. So the added storage and processing and reduced p2 time of 1080 werent justified - we were thrilled with the quality of 720.

Make sure you need 1080 if you use it - if you dont need it you will be swapping cards more then you need to and peforming backups and backups more then you need to - the on set workflow is not trivial.

see my report on the filmout here
http://www.dvxuser.com/V6/showthread.php?t=51195&highlight=filmout

and you can see trailers from the film here (uncolor corrected) - http://www.americanstandardthemovie.com/media.html

Thanks for the link... some interesting info there, especially about print versus camera stocks for film-out.

Nothin' to do at this point except test!