Tim: 720p60 and 24p questions

[Steve Mullen]

Tim wrote:

"The nice thing is that (in FCP) if you edit 24P footage from the source 720P60 clip into a 23.98 fps sequence, the repeat pulldown frames are automatically thrown away - no matter what frame you start on."

How does this happen?

The camcorder outputs 720p60 which is converted to HD-SDI which is then digitized into a stream of 60fps. There are no flags. But, there is a 2:3 pulldown cadence in the 59.94 video.

When you move one of these clips into a 23.98fps Sequence -- how does FCP know "what frames should be dropped?"

You are right that you can move these clips in a 23.98 Sequence, but I can't see how FCP knows which frames are from RFs.

Is FCP smart enough to discover -- like AE -- the 2:3 cadence and move only the first of every "2" and the first of every "3" into the Timeline?

Or, do you mean you manually trim the start of each clip to the first of the "2" so that one gets the so-called A frame?

Assume that's what you do -- what happens when the clip is moved into the Timeline?

1) Is FCP smart enough to move only the first of every "2" and the first of every "3"?

2) Does FCP simply play the first of every "2" and the first of every "3"?

I suspect the latter.

------------

Interesting that HDVxDV outputs AIC at 1280x1080 and DVCPRO HD at 960x1080 -- neither of which is correct. You must Preferences to "HDTV at 1280x720" -- that outputs at 960x720. At least this is correct. And, AIC is also correctly output at 1280x720.

-------------

Now that we know that AIC does not lose quality with progressive video, it seems like it is ideal for output from HDVxDV rather than DVCPRO HD which causes a loss of H. rez.

Both are output, as you noted, at 24fps not 23.98.

In reality, there are 24 frames in every second. Right?

When dropped in either a 23.98 or 24 Timeline -- the full number of trimmed frames is dropped into the Timeline.

So here are 2 questions:

1) When the Sequence is played, are the frames clocked-out at 23.98 or 24? If at 23.98, then the CLIP timebase tag of 24 has no relevance.

2) It would seem that 23.98 should be Drop Frame -- yet that option is not allowed even When the Sequence timebase is 23.98. Curious?

[Tim Dashwood]

Originally Posted by Steve Mullen

Tim wrote:

"The nice thing is that (in FCP) if you edit 24P footage from the source 720P60 clip into a 23.98 fps sequence, the repeat pulldown frames are automatically thrown away - no matter what frame you start on."

How does this happen?

The camcorder outputs 720p60 which is converted to HD-SDI which is then digitized into a stream of 60fps. There are no flags. But, there is a 2:3 pulldown cadence in the 59.94 video.

When you move one of these clips into a 23.98fps Sequence -- how does FCP know "what frames should be dropped?"

You are right that you can move these clips in a 23.98 Sequence, but I can't see how FCP knows which frames are from RFs.

Is FCP smart enough to discover -- like AE -- the 2:3 cadence and move only the first of every "2" and the first of every "3" into the Timeline?

Or, do you mean you manually trim the start of each clip to the first of the "2" so that one gets the so-called A frame?

Assume that's what you do -- what happens when the clip is moved into the Timeline?

1) Is FCP smart enough to move only the first of every "2" and the first of every "3"?

2) Does FCP simply play the first of every "2" and the first of every "3"?

I suspect the latter.

FCP isn't doing anything special when you put a 23.98P @ 59.94 clip into a 23.98fps sequence. It just does what it has always done in the resolution/frame rate independent environment FCP works in.
Unlike Avid (except for some of the newer flavours) FCP doesn't require the source clip to be the same resolution/frame rate/codec as the sequence.
Therefore you can drop a PAL clip into a NTSC sequence and FCP will automatically add the appropriate pulldown frames. You can drop a NTSC clip into a PAL sequence, and FCP will automatically drop the appropriate amount of frames and maintain "real" time. This also works for dropping NTSC or PAL into a HD sequence of any frame rate, size or codec.

The only problem with letting FCP handle PAL to NTSC to 24P conversions is that the math is simple. It just drops or adds the appropriate amount of frames/fields to make it work. That's why plugins like Graeme Nattress' or Magic Bullet's are so important for "intelligently" creating film-looks from interlaced video.

So that brings me to the math of 24P within 60P. This really is the simplest pulldown math possible. No ABC or D frames or fields to worry about - just an extra frame added every fifth. This means that no matter what frame you start on, the "simple math" calls for FCP to just drop 36 of the 60 frames - and make it even. Hence 2:3 pulldown.
It just does what comes natural.


------------

Originally Posted by Steve Mullen

Interesting that HDVxDV outputs AIC at 1280x1080 and DVCPRO HD at 960x1080 -- neither of which is correct. You must Preferences to "HDTV at 1280x720" -- that outputs at 960x720. At least this is correct. And, AIC is also correctly output at 1280x720.

Yes. I've always just used my own custom settings. It is actually possible to trick the DVCPROHD codec into 1280x720 by using a custom size - but I prefer uncompressed or AIC. I also batch conform the new clips to 23.98 just to make sure there will be no sync problems.

-------------

Originally Posted by Steve Mullen

Now that we know that AIC does not lose quality with progressive video, it seems like it is ideal for output from HDVxDV rather than DVCPRO HD which causes a loss of H. rez.

And no one believed me when I said a month ago I saw no added artifacts or loss in colour! I think the bad reports were from version 1, but the current version of the codec is 1.0.1.

Originally Posted by Steve Mullen

Both are output, as you noted, at 24fps not 23.98.

In reality, there are 24 frames in every second. Right?

Only in film aquisition and projection or HDCAM are there actually 24fps. Even a teleciné transfer of film cannot run at 24fps, it must run at a slower 23.98 to be put on videotape. This is why when you shoot a music video on film, you must use a digitach and set it to 23.98 - or you will eventually lose sync on very long takes.

However, Avid Symphony or Avid film composer & FCP have the ability to actually run the footage at true 24fps after it has been digitized. This is what you would want if you were cutting a feature film and wanted to make sure the sound was running at the right speed/pitch for the final mix and optical print. Although you still wouldn't be able to lay it back to videotape with TC at 24fps, it would need to be 23.98.

Originally Posted by Steve Mullen

When dropped in either a 23.98 or 24 Timeline -- the full number of trimmed frames is dropped into the Timeline.

So here are 2 questions:

1) When the Sequence is played, are the frames clocked-out at 23.98 or 24? If at 23.98, then the CLIP timebase tag of 24 has no relevance.

The frame are clocked out at whatever the sequence is actually set to - 23.98 or 24. You may find that you can actually drop 24fps clips into 23.98 sequences (and vise-versa) without any rendering or transcoding. I think FCP is aware that for editing purposes they are the same thing. The frame rate setting of the sequence is what matters.

Originally Posted by Steve Mullen

2) It would seem that 23.98 should be Drop Frame -- yet that option is not allowed even When the Sequence timebase is 23.98. Curious?

No film transfer or 24P camera can record drop frame TC because the simple math wouldn't work for the keycode matchback. Only 60i masters prepared for broadcast need to ever be in drop-frame so that you know actual run-length.

Whether TC is in DF or NDF, NTSC video is always consistently running at 29.97 fps. DF TC is just a means of compensating so that we can count the runtime minutes in a hour properly for broadcast.
Since NTSC is always 29.97, then a 2:3 pulldown sequence must run at 23.976fps (aka 23.98)

29.97/30=0.999
23.976/24=0.999
59.94/60=0.999

I think where JVC may have screwed up is in TC for 24P. They claim that the camera uses SMPTE TC, but upon closer inspection they use 24fps TC when in 24 mode. (:00 to :23) This is very weird - I think only HDCAM works this way. I would have expected 60fps TC or maybe even true SMPTE.
This is probably the major reason why all NLEs are having so many problems with capturing 720P24 material from the HD100. There are no existing control templates for 24fps TC in HDV. It is brand new and will require some re-thinking on the part of Apple and Avid.

[David Dessel]

Tim,

I would just like to confirm the exact workflow that can be interpreted from this thread.

1.) Using HDVxDV, convert the mt2's to AIC 1280 X 720.

2.) Using Cinema Tools, Batch Conform the 24fps files to 23.98. (NOTE I COULD NOT GET THIS STEP TO WORK.)

3.) Create a FCP Project using the 720-30P AIC Preset.

4.) Import Clips.

5.) Create a 23.98 Sequence for editing.

Is this correct?

Also, my apologies for not being able provide feedback on the presets you E-Mailed me. I will do so as soon as me schedule opens up. In any case, thank you very much for sending them.

Best Regards,

Dave Dessel

[Tim Dashwood]

Originally Posted by David Dessel

Tim,

I would just like to confirm the exact workflow that can be interpreted from this thread.

1.) Using HDVxDV, convert the mt2's to AIC 1280 X 720. Yes - but also use 48Khz 16 bit for sound

2.) Using Cinema Tools, Batch Conform the 24fps files to 23.98. Open Cinema Tools, hit cancel when it asks you to open a database file, select File-Batch Conform, select one of your new clips created with HDVxDV, select 23.98. That's it.

3.) Create a FCP Project using the 720-30P AIC Preset. The preset isn't necessary as we will create a custom sequence setup.

4.) Import Clips. yes

5.) Create a 23.98 Sequence for editing. Yes, but make a custom sequence preset once for AIC 720P24 with the following settings: 1280x720 HDTV 720P (16:9), pixel aspect SQUARE, editing timebase 23.98, Timecode rate 24, Quicktime video settings AIC compressor, compressor preset to HDV 720P, audio 48Khz, 16-bit.
[/QUOTE]

[Steve Mullen]

The current HDV@Work describes existing four work flows. You can find the HDVxDV flow in detail at:

www.mindspring.com/~d-v-c

[Steve Mullen]

Originally Posted by Tim Dashwood

So that brings me to the math of 24P within 60P. This really is the simplest pulldown math possible. No ABC or D frames or fields to worry about -- just an extra frame added every fifth. This means that no matter what frame you start on, the "simple math" calls for FCP to just drop 36 of the 60 frames - and make it even. Hence 2:3 pulldown. It just does what comes natural.

Tim, I'm really confused by this.

When you say "drop" you mean "don't play" the frame, right?

Do you mean it plays 1, skips 1, plays 1, skips 2, etc.

AABBBAABBBAABBB
A_B_ _A_B_ _A_B_ _

it seems to work if AA are first

ABBBAABBBAABBB
A_B_ _A_B_ _A_B_ _

it also seems to work if AB are first

BBBAABBBAABBB
B_B_ _B_B_ _B_ _

BBAABBBAABBB
B_A_ _B_A_ _B_ _

BAABBBAABBB
B_A_ _B_A_ _B_ _


but not if it starts with B. It seems you must trim so the first frame is an A, either the first or second.

[Steve Mullen]

Originally Posted by Tim Dashwood

This is why when you shoot a music video on film, you must use a digitach and set it to 23.98 - or you will eventually lose sync on very long takes.

Although you still wouldn't be able to lay it back to videotape with TC at 24fps, it would need to be 23.98.

You may find that you can actually drop 24fps clips into 23.98 sequences (and vise-versa) without any rendering or transcoding. I think FCP is aware that for editing purposes they are the same thing. The frame rate setting of the sequence is what matters.

If you create a 23.98 Sequence in FCP according what you say, it will be clocked-out at 23.98fps. I think you are right.

And yes, you can drop 24 into 23.98 without any need to render.

Since you cannot record either 23.98 or 24 to tape -- unless there is some special deck -- one has to get the 23.98 to 29.97. Which means adding 2:3 pulldown. Now FCP has such a function that was used for the DVX100. (Don't know if it works with HD, however.)

That leads me to feel all is fine -- except for your comment about loss of sync. If you recorded a clock for an hour at 23.98 to HDV -- then converted it to AIC or DVCPRO HD with a timebase wrongly set to 24 and then dropped this into a 23.98 Sequence that ran at 23.98 -- would there really be any loss of sync? I don't think so as the 24 tag has been ignored.

[Tommy James]

If you are going the HD-SDI route why not simply shoot in the 24p mode which outputs 48p in analog which is then digitized with the HD-SDI card and then compressed using the Cineform Prospect HD ingest software and captured using your Dual Opteron workstation. The 48p would make good 24p slomo or every other frame can be dropped. Or 48p can be output to 60p using 1112 pulldown. With the advent of digital cinematography 24p can be used for the drama while 48p will be used for the fast action in a hybrid format.

[Tim Dashwood]

Originally Posted by Steve Mullen

Tim, I'm really confused by this.

When you say "drop" you mean "don't play" the frame, right?

Yes. Drop-frame TC is something completely different.

Originally Posted by Steve Mullen

Do you mean it plays 1, skips 1, plays 1, skips 2, etc.

Yes.

Originally Posted by Steve Mullen

but not if it starts with B. It seems you must trim so the first frame is an A, either the first or second.

On paper it seems that starting on the first "B" frame would not work and end up skipping every other frame. However, I just performed a test starting at all 5 possible frames, and it still works! I can't explain it - but it works. It seems to treat the first B as if it were a A.
Here is my test: http://homepage.mac.com/timdashwood/...60to24test.mp4

[Tim Dashwood]

Originally Posted by Steve Mullen

Since you cannot record either 23.98 or 24 to tape -- unless there is some special deck -- one has to get the 23.98 to 29.97. Which means adding 2:3 pulldown. Now FCP has such a function that was used for the DVX100. (Don't know if it works with HD, however.)

It all works the same way. Realtime playback of 24P to NTSC.

Originally Posted by Steve Mullen

That leads me to feel all is fine -- except for your comment about loss of sync. If you recorded a clock for an hour at 23.98 to HDV -- then converted it to AIC or DVCPRO HD with a timebase wrongly set to 24 and then dropped this into a 23.98 Sequence that ran at 23.98 -- would there really be any loss of sync? I don't think so as the 24 tag has been ignored.

That would be fine assuming you were recording sync sound to tape. It should always maintain sync with the picture it is attached to.

However, when I was explaining the difference between 24fps and 23.98 earlier, I was referring to double-system where a Nagra or DAT is used to record crystal sync sound. If you shoot 24fps on film, transfer it via teleciné to videotape (23.98), digitize it and the sound from its source (1/4" Nagra, digital Nagra, or DAT) then the picture would be running slightly slower than the sound. This is when you would need to change the sequence to 24fps - and then everything will be back to running at the same speed as when the film was shot.

In most cases there would be enough budget to hire the lab to prepare sync dailies instead of the assistant editor doing all the work. The lab would sync the double-system at the corrected rate, which would therefore always be in sync with the picture.

[Tim Dashwood]

Originally Posted by Tommy James

If you are going the HD-SDI route why not simply shoot in the 24p mode which outputs 48p in analog which is then digitized with the HD-SDI card and then compressed using the Cineform Prospect HD ingest software and captured using your Dual Opteron workstation. The 48p would make good 24p slomo or every other frame can be dropped. Or 48p can be output to 60p using 1112 pulldown. With the advent of digital cinematography 24p can be used for the drama while 48p will be used for the fast action in a hybrid format.

Tommy,

The HD100 doesn't have HD-SDI out, but does have analog Component out.
THe 48fps output only works when capturing "live" from the output. You cannot get 48fps or 60fps out of the analog out during playback from tape.

I know you mean well, but this specific thread is in reference to research being performed on Final Cut Pro on Mac, not Cineform on Windows.
However, we are aware of the great products Cineform is offering, and your suggestion of using the higher frame rates for "overcranked" slo-mo is something I will be researching this week.

[Steve Mullen]

Originally Posted by Tim Dashwood

However, I just performed a test starting at all 5 possible frames, and it still works! I can't explain it - but it works. It seems to treat the first B as if it were a A.
Here is my test: http://homepage.mac.com/timdashwood/...60to24test.mp4

You are right. I captured the first frame of each segment and lined then up vertically so I could see the boy's hair move. It steps forward exactly as it should.

Looks like when a clip is dropped into a 23.98 Sequence a quick analysis is done over the first frames to determine the B frames -- three identical frames -- and tags the clip appropriately. Now the playback system can reference the tag so it always knows what frames to play and which to skip. Very clever.

I wonder if it will work with multiple clips and if you trim them in the timeline?

-----------------------

"It all works the same way. Realtime playback of 24P to NTSC."

Which means the 720p23.98 should become 720p29.97. That should convert to NTSC fine. It should also be able to be exported to 720p30 HDV as well. Of course, it will have a "telecined filmlook."

----------------

"That would be fine assuming you were recording sync sound to tape. It should always maintain sync with the picture it is attached to."

Coming from the HD100 that would be the case. If you are concerned about MP2 -- someone needs to try the PCM track.

Someone with a mixer that outputs a LINE-LEVEL signal to an XLR plug might want to input it to Channel 1.

Be sure the MIC/LINE switch is set to LINE.

If it works one would have a double-system approach using the HD100 itself!

[Alex Humphrey]

OK.. understood all of the above. Working with JVC HD110 and FCP Studio and experimenting with the trial version of HDVxDV program. I'm looking to take SD-HDV 960x480 60fps from the JVC, captured with HDVxDV (trial version works fine for testing) export to AIC 60fps. (I did DVCPROHD 720p 60fps before work this morning, I'll try AIC after work tonight)

but taking the DVCPROHD 720p 60fps and dropping it into a 24fps or 30fps timeline.... FCP Studio brings in only the frames it needs to conform to the timeline. I'm going to guess it would make more sense to make another project in a 60fps timeline.. change the speed to 40% (24fps) or 50% (30fps), then rexport that file to a new clip to be dropped into the 24fps or 30fps project that I'm working on? I didn't figure out a faster way this morning before coffee, so I might easily be missing something very basic.