Edit points on 23.98 timeline

[Joseph Owens]

[Dave LaRonde] “the conversion is simple: 3 frames, then 2 frames, then 3 frames, then 2 frames and so on.”

Wrong.

This is where all this disinformation gets passed along.

3:2 cadencing occurs at the *field* level, not frames. I don’t think Apple actually understands this, either, because Final Cut doesn’t do it right, and you certainly cannot drop a 23.98 edit onto a 29.97 timeline without it literally exploding. You can’t even export a properly cadenced Quicktime from FCP. Its really quite pathetic and amateurish.
They can’t even include 23.98 as a major rate in their “Easy Setups”.

This is how it works:

24 film frames, 30 video frames (would be easy, but as we all know, NTSC video frame rate is actually 59.94 interlaced fields per sec. So we pull down the film frame rate to 23.9759 fps.

Then we generate redundant fields so that 4 film frames occupy five video frames (= 10 video fields), in the pattern 2:3:2:3 fields. There is a way of describing the film frame sequence identifying the progression as ABCD. An “A” frame is the case where a single film frame occupies a single video frame (contiguous fields 1,2). The next 3-field sequence is a “B” frame, which is video frame 2 plus the first field of frame 3. A “C” frame, which is the one we want to avoid, is field 2 of the third video frame and field 1 of the fourth video frame. You can see why this one causes a lot of problems. Ending the cycle is the “D” frame which is field 2 of video frame 4 and both fields of the fifth frame. And just to be clear, these are frames 00, 01, 02, 03 and 04, in NON-drop.

It is as elegant as 1950’s electronics can be. It depends on interlacing. It is very difficult to see in the Final Cut timeline as FCP does not actually work in fields plus all displays these days are progressive and all you ever hear is the whining about the juddering and shaking and twittering. I have a crappy CRT sitting downstream just so I can see this. And you’d be left breathless with wonder at how often FCP messes it up. Think of another word for “messes” if you want to know what I think of Apple’s fudged kluges.
Panasonic came up with Advanced Pulldown, which is 2:3:3:2, actually an ABDA sequence, which does away with the problematic “C” frame, as all you have to do, to “remove the pulldown”, is throw away the middle video frame as it is a mixed-field frame containing information which already exists in the adjoining frames.

When you are working in Non-drop frame, it is easy to keep track of the A-frame sequence, especially if your crack telecine operator has done a proper A-Frame transfer (I’m harking back to actual film days)
and hit A’s on 00, 05, 10, 15, 20, 25 — and you can avoid C’s. Drop frame, however, skips forward in its count and the upshot is that the A-frame-to x5 correlation is lost after the first minute.

SMPTE spec is to start counting at zero, but with most FCP timelines beginning at DF 1:00:00;00, that actually works out to the first frame being a “C” frame. How’s that for unfortunate?

As a long-time telecine colorist, it is something that became ingrained with me in day-to-day operations, but I can understand how most video editors, who have recently had to start dealing with how to integrate different frame rates have not yet come to terms with this fundamental concept. But those days are past.

jPo

You mean “Old Ben”? Ben Kenobi?

[Shane Ross]

No. I’ve done more 23.98 projects than I have fingers and toes…you don’t need to worry about that.

Shane
Little Frog Post
Read my blog, Little Frog in High Def

[Joseph Owens]

So I won’t tell anyone about the “hook hand”… ;P

jPo

You mean “Old Ben”? Ben Kenobi?

[Shane Ross]

Or my peg leg…
😉

Shane
Little Frog Post
Read my blog, Little Frog in High Def

[Jack Avalon]

Thanks for the input chaps. I suspected that it was not an issue since I searched the forums and never found anything on the subject, but I have always fretted a bit when I see those split frames after a cross/down conversion from 23.98.

Jack Avalon
Third House Productions

[John Christie]

Jack

The key point here is how are you going to add pulldown? FCP won’t do a proper job by itself. You need a video card like a Kona or Blackmagic to add the correct pulldown to a 23.98 sequence while outputting to 29.97. We do this all the time, edit in 23.98 and output to 29.97, mostly using the Kona 3. Looks great.

Cheers

[Doug Beal]

If you do not fix frames that have interfield motion (the results of 23.98 edit landing on field2 of a 29.97 conversion it will get kicked back by some companies BitMax for one.
we have a template overlay that can drop on top of a 23.98 edit that shows where edits can occur that will not result in interfield motion once the pulldown is applied.

We print to HDCam at 23.98 and reinjest at 1080i with the HDCam machine generating the pulldown with both A frame 24 and A frame 30 being set to 01:00:00:00. This allows TC to be the same every second on the second and to generate the cadence from 1 hr forward.

Yet another reason to use a broadcast monitor that can show both fields as a frame and a reason to keep the Kona cards set to Pause = full Frame.
The bad edit won’t display on a plasma or in FCP canvas window.

Doug Beal
Editor / Engineer
Rock Creative Images
Nashville TN

[Joseph Owens]

[Dave LaRonde] “1080 has 59.94 FIELDS per second. 720 has 59.94 FRAMES per second. Big difference.”

Ah, yes ,the ABC network.

I guess its a matter of semantics whether 1/60th of a second is a progressive entity or half an interlaced 29.97 frame. So, how do the at-home viewers interpret 720P59.94 on their NTSC receivers?

jPo

You mean “Old Ben”? Ben Kenobi?

[John Heagy]

Discreet products are the only place I’ve seen indicators telling one an edit would result in a split cut once 3:2 was applied.

[Matt Lyon]

Interesting discussion!

I’ve never honestly even thought this might be an issue. Nor has anyone ever told me this could be an issue. I’ve always just edited at 23.976, exported w/ 3:2 and been done.

[Doug Beal] “If you do not fix frames that have interfield motion (the results of 23.98 edit landing on field2 of a 29.97 conversion it will get kicked back by some companies BitMax for one.”

Is this a matter of their QC machines being overly conservative? Have you ever had an actual broadcaster reject a master because of an edit landing on a split field? Not saying it doesn’t happen, I’ve just never seen this on a QC report. But I’m not an online guy, so I don’t see that many QC reports 🙂

Matt Lyon
Editor
Toronto