A Few Basic Questions: 24pA, Scene Files

[righteous.]

I've mostly been shooting 30p on my DVX100b, but my new goals as a filmer are to shoot almost everything at 24p because, in my mind, nothing beats a well shot sequence at regular speed.

1. Anyhow, I realize that I'd have to do pulldown removal/inverse telecine for 24p footage, and I don't batch capture so I REALLY can't take the time to separate the clips. If I understand this correctly, if I shoot 24pA, does this mean I won't have to do pulldown removal? And 24pA gets put into a 24 fps timeline, but what about the 30p that I mix? Does this mean I'll have to do pulldown removal to the 30p footage in order to integrate? Will I still be able to slow it down 80% to get a progressive framerate the way I do in a 29.97fps timeline?

2. I have been shooting with my master pedestal at -8 for years because thats the consensus amongst other DVX filmers I knew. When I filmed skiing, the trees would always be pitch black unless I completely overexposed the snow. I upped it to -2 and it looks better. My question is: does master pedestal only effect lows? Or is it more of a low/mid calibration?

3. Does cinelike_D really have the greatest dynamic range? From what I can tell looking at the LCD, it looks like the regular Cinelike gamma setting displays snow detail better. Is there something I'm missing?

4. Which V_Detail setting should I shoot for display on a computer monitor? I am at college and left my DVX Book at home. I can't remember if the THIN or THICK setting was for monitors...

 

[David Jimerson]

1) 24p vs 24pA -- only a capture concern if using FCP. Everything else handles them the same as each other.

Yes, you will have to remove pulldown with 24pA. There's a myth out there that it's recorded to tape as straight 24p, but it's not -- it's recorded in a 60i stream just like 24p standard; the only difference is how the pulldown is arranged.

If you're going to mix 24p and 30p, shoot 24p standard (not 24pA) and edit on a 60i timeline (DV NTSC). 24p and 30p do not convert to each other well, but both do fine in 60i.

Unless, of course, you slow the 30p down to 80% for slo-mo, like you're saying. Then, you do that on a 24p timeline.

Not sure what this means:

Will I still be able to slow it down 80% to get a progressive framerate the way I do in a 29.97fps timeline?

Whether or not something is progressive doesn't have anything to with whether it's slowed down or speeded up; it's whether or not whole pictures instead of half-picture fields were shot. If footage is progressive, you don't have to change its speed to make it that way.

2) I think most of us like our master pedestal at about -3 to -5, depending on the gamma curve. But what you're seeing with the trees and the snow has to do with the dynamic range of the camera -- the width of brightness the camera is able to resolve. The master pedestal more or less affects the midpoint of it, but it doesn't change the range. So, if you turned it way up, you'd see something of an opposite effect.

3) Cinelike D is meant for stretching the dynamic range, yes.

4) THIN. Preserves the most vertical detail.

 

[righteous.]

1) 24p vs 24pA -- only a capture concern if using FCP. Everything else handles them the same as each other.

Yes, you will have to remove pulldown with 24pA. There's a myth out there that it's recorded to tape as straight 24p, but it's not -- it's recorded in a 60i stream just like 24p standard; the only difference is how the pulldown is arranged.

Does this mean I'll have to remove pulldown using a different method? Or can I just use the "inverse telecine" button in Cinema Tools?

If you're going to mix 24p and 30p, shoot 24p standard (not 24pA) and edit on a 60i timeline (DV NTSC). 24p and 30p do not convert to each other well, but both do fine in 60i.

Could this mean that my interlacing artifacts on my 24p footage are due to the fact that I'm always working on a 29.97 fps timeline? I don't think I've ever worked in 60i...

Unless, of course, you slow the 30p down to 80% for slo-mo, like you're saying. Then, you do that on a 24p timeline.

Not sure what this means:

Whether or not something is progressive doesn't have anything to with whether it's slowed down or speeded up; it's whether or not whole pictures instead of half-picture fields were shot. If footage is progressive, you don't have to change its speed to make it that way.

Wouldn't the 30p footage show interlacing artifacts if it were on a 24p timeline? Does the 80% slow down trick work the same on a 24fps timeline? I thought I read in Barry's Book that shooting 30p at 1/120 slowed down 80% on a 29.97fps timeline had some sort of effect where everything "added up right," and made a pseudo-progressive slow motion framerate? I know that if I slow down my 30p footage by any percent, I have to render in order to play it back. But if I do it 80%, there is no rendering needed...

2)

I think most of us like our master pedestal at about -3 to -5, depending on the gamma curve. But what you're seeing with the trees and the snow has to do with the dynamic range of the camera -- the width of brightness the camera is able to resolve.

Does this mean that the effect of the master pedestal on the image is directly related to the gammer curve? For example, a -5 on one gamma setting may have more crushed blacks than a -5 on another setting, correct?

The master pedestal more or less affects the midpoint of it, but it doesn't change the range. So, if you turned it way up, you'd see something of an opposite effect.

Maybe its just the wording, but I don't know exactly what you mean by this? could you elaborate please?

 

[David Jimerson]

Does this mean I'll have to remove pulldown using a different method? Or can I just use the "inverse telecine" button in Cinema Tools?

If you shot 24p standard, then the only thing you can do is use IVTC in Cinema Tools (if using FCP). If you're planning to edit as 23.976 (which you should unless you're mixing frame rates without slo-mo), then always shoot 24pA.

Could this mean that my interlacing artifacts on my 24p footage are due to the fact that I'm always working on a 29.97 fps timeline? I don't think I've ever worked in 60i...

You will always see interlace artifacts on a 60i timeline (which IS an NTSC DV "29.97" interlaced timeline.) 24p footage must have interlaced fields inserted in it in order to comform to the frame rate (which is what pulldown is).

You won't, however, see them on a standard TV.

Wouldn't the 30p footage show interlacing artifacts if it were on a 24p timeline?

Depends. Not inherently, because there's no interlaced footage involved, in theory. But it could if the NLE is treating it as straight NTSC footage.

Does the 80% slow down trick work the same on a 24fps timeline?

A 24p timeline is where it works ideally, because 24 is 80% of 30 and each 30p frame then lines up with a 24p frame, meaning that no frame will have to be resampled. Each 30p frame just becomes a 24p frame, but because it's being played at 24 fps instead of 30, it plays back slower.

I thought I read in Barry's Book that shooting 30p at 1/120 slowed down 80% on a 29.97fps timeline had some sort of effect where everything "added up right," and made a pseudo-progressive slow motion framerate?

He was referring to a 24p timeline.

I know that if I slow down my 30p footage by any percent, I have to render in order to play it back. But if I do it 80%, there is no rendering needed...

Any time the math works out to whole numbers, it's probably fine. If the footage doesn't have to be resampled, then it wouldn't have to be timeline-rendered.

Does this mean that the effect of the master pedestal on the image is directly related to the gammer curve? For example, a -5 on one gamma setting may have more crushed blacks than a -5 on another setting, correct?

Yes and no. Yes, because it will appear to have a deeper effect, but no, because it does the same thing regardless; it's just starting with a different image.

I dunno -- there could be a better analogy, but it's like adding hot cocoa mix to water or to milk. You do the same thing to both, but the results are different because of what you start with.

Maybe its just the wording, but I don't know exactly what you mean by this? could you elaborate please?

Hard to explain in words. If you think of the master pedestal as a "fulcrum," where you place the fulcrum affects what a lever does. Think of the gamma curve as a lever, with the highs and lows of the dynamic range as the end points.

 

[righteous.]

If you shot 24p standard, then the only thing you can do is use IVTC in Cinema Tools. If you're planning to edit as 23.976 (which you should unless you're mixing frame rates without slo-mo), then always shoot 24pA.


You will always see interlace artifacts on a 60i timeline (which IS an NTSC DV "29.97" interlaced timeline.) 24p footage must have interlaced fields inserted in it in order to comform to the frame rate (which is what pulldown is).

You won't, however, see them on a standard TV.

Depends. Not inherently, because there's no interlaced footage involved, in theory. But it could if the NLE is treating it as straight NTSC footage.

A 24p timeline is where it works ideally, because 24 is 80% of 30 and each 30p frame then lines up with a 24p frame, meaning that no frame will have to be resampled. Each 30p frame just becomes a 24p frame, but because it's being played at 24 fps instead of 30, it plays back slower.


He was referring to a 24p timeline.


Any time the math works out to whole numbers, it's probably fine. If the footage doesn't have to be resampled, then it wouldn't have to be timeline-rendered.

So what you're saying is that playing 30p footage on a 29.97fps timeline at 80% is the same as playing 30p footage at regular speed on a 23.976 fps timeline?

By the way, I never knew the 29.97 was an interlaced timeline. I assumed it was 30 fps progressive. Does this mean that 23.976fps is interlaced somehow? how exactly is this different from a 24fps timeline? I realize that the number 24 is an approximation, but is that ALL it is? Basically I should work in actual values (23.976/29.97) instead of approximations? (24/30)

Yes and no. Yes, because it will appear to have a deeper effect, but no, because it does the same thing regardless; it's just starting with a different image.

I dunno -- there could be a better analogy, but it's like adding hot cocoa mix to water or to milk. You do the same thing to both, but the results are different because of what you start with.

That makes sense. My question is: which comes first? Does the gamma curve play off the master ped, or vice versa?

 

[David Jimerson]

So what you're saying is that playing 30p footage on a 29.97fps timeline at 80% is the same as playing 30p footage at regular speed on a 23.976 fps timeline?

No; I'm talking about playback speeds relative to the footage and how it matches up to the timeline you're working on.

If you're playing back 30 fps footage at 80% of its native rate, you're playing it back at 24 fps. This matches up to a 24 fps timeline exactly, so whole frames match up. You don't have to discard, insert, or blend any frames make it match.

Any time the math works out that the number of pictures taken per second (24 for 24p, 30 for 30p, or 60 for 60i), when you factor in the speed change, matches up in whole numbers to the timeline you're working on, you get best results.

So, if you take 60i footage and slow it down to 40%, you have .4 x 60, which equals 24. So, every interlaced fields matches up with a 24p frame, and you get smooth motion because you don't have skip, add, or blend any frames or fields. (Now, because they're half-pictures, fields, you have to fill them in to make them whole frames, but that affects apparent vertical resolution, and not motion smoothness.)

And yes, these numbers are actually 29.97, 23.976, and 59.94, but it's easier to understand them using 30, 24, and 60. The math all works out to the same, because they're derived from the same formula (30 x 1000/1001; 24 x 1000/1001; 60 x 1000/1001).

By the way, I never knew the 29.97 was an interlaced timeline. I assumed it was 30 fps progressive.

An NTSC timeline is "29.97" and interlaced, but not every 29.97 timeline is. Check the field order -- if it's "upper field first," then it's interlaced. If it's "none" or "progressive," then it's progressive. (In PAL, interlaced footage is "lower field first.")

But NTSC is commonly called "29.97 fps" -- this is because of some archaic and technically inaccurate legacy nomenclature. It would also go away if there were no more interlaced video of any kind.

The simple thing to remember, especially when it comes to thinks like slo-mo, is the number of times per second a picture is captured. For 24p, that's 24 times. For 30p, that's 30 times. For 60i, that's 60, not 30 times. It's 60 half-pictures, but it's 60 moments in time nonetheless, not 30. And that's the reason you get interlace artifacts at all, because moving objects are two different places between the fields, and if you try to shove them together to make a "frame," it's exactly like slicing up two photographs, slapping the halves together, and calling them a single photo. Anyone can see they're not.

But the idea of "29.97 frames per second" persists, because it was around for so long. You can call it a "frame," just like you can call a lamb's tail a "leg," but it's actually two half-frames.

Does this mean that 23.976fps is interlaced somehow?

No. There is no such thing as an interlaced 23.976 timeline, nor is any footage shot that way. (I suppose you could make one, but there's no reason on Earth to do so.)

The only way 24p footage ever becomes "interlaced" is when pulldown -- extra fields -- is inserted to conform it to 60i. But it's only "interlaced" as long as it's in that 60i stream (and it still looks like native 24p); once you remove pulldown, it again becomes 24 progressive frames.

That makes sense. My question is: which comes first? Does the gamma curve play off the master ped, or vice versa?

I don't honestly know which happens first in the DSP, but because you can't change the order, it doesn't really matter for practical purposes -- and it might not matter at all, anyway. At least not in every case.

But you can play with this in post by taking a flat image and applying RGB color curves and changing the black offset, and changing the effect order.

 

[righteous.]

Makes sense!

But now that I try to do Inverse Telecine in Cinema Tools, it asks what capture mode my clips is, and my options say: Field 1 only, Field 2 only, F1 - F2, F2 - F1, amongst other things. How am I supposed to know any of this information? When I do the INTC it still has the lines...

 

[David Jimerson]

You want F1-F2. If you're starting with the very first frame in the clip, you want to start with Frame A.

For really fine-tuned advice, ask the learned souls in the FCP section; stuff like this is why I don't use it.