Frequently Asked Questions
Currently working on an HD feature in NY using the Panasonic
Varicam. We're shooting at 24P and I am reading about the difference between 24 and 23.98 and need a little help. I spoke with the rental house and it seems like they can't tell me why.
Our camera is in 24P mode, with the deck recording at 29.98 NDF. Our sound is being fed via XLR from a boom mic though a mixer and into the camera. I'm not terribly concerned about what the sound dept. has, which you'll understand in a moment why.
It seems as if the sound is approx. two to three frames ahead of the video on the original master tapes. I asked the editor if it was possibly something wrong with the avid or in her post chain somewhere, but the post house used a deck and a pair of headphones with an HD monitor straight out of the deck. I would have a few theories if it was behind the video, but having the sound precede the video has got me
scratchin' my head. We've even switched out the body, hoping it was some flaw in the processing or sampling in the camera itself. Ideas? Any help would be greatly appreciated.
|Here are three guesses: 1. The audio is not being delayed to compensate for the HD compression delay. In
HDCam, I know the compression takes a full frame to compress and also to decompress. The decks and cameras have to delay the audio on both recording and playback to compensate for this delay... I would assume that
DVCPRO-HD is the same, but if there are no audio delays, the audio would be ahead by two frames by the time it plays out.
2. I could see audio being two frames ahead at the post house if camera master is being played out upconverted to 1080p/24 for viewing on the HD monitor at the post house. The audio from the headphones may not be delayed approprately to compensate for the upconversion process. 3. The Avid is using
downconverts, the audio may not have been transfered properly (compensating for the delay in downconverting the video). The Avid editor looks at the timeline and sees a certain delay, then the person at the post house on headphones is told to look and see if there is a two frame audio advance and it becomes a case either #2 above, or "you hear what you are told is there"... I know it has happened to me. ... beyond that???
Under what circumstances would one use 24p versus 23.98p? I have checked out the Cinealta 24p camera and found out that it offers 24p as well as 23.98p frame rates. Panasonic's Varicam offers straight 24p and not 23.98p, I believe but am not sure. Panasonic's
AG-DVX100 mini DV 24p "a format that has nothing to do with
HD" camcorder offers a 24p and a 24p enhanced mode.
My understanding is that with the F900 Cinealta the 23.98p is if the production is never to go to actual film in the final product and only stay in videoland; whereas an actual film output would necessitate true 24p shooting. Is this correct thinking?
If the choice between 24p and 23.98p is critical, this makes me wonder whether or not a 24p
mini "a format that has nothing to do with HD" camcorder will have issues in post.
I am curious to know the nuances of this framerate choice so that I can speak from an informed position with a director and producer and not speculate and "hope for the best" when it goes down the line to post-production.
Also, do editing programs like Avid and Final Cut Pro handle 24p or 23.98p interchangeably?
I'm sorry if this is too basic, but it really is the basics that are the root of the answer: Like it or not, the
frame rate of NTSC is 30/1.001 fps or roughly 29.97fps. Film cameras and projectors have traditionally run at exactly 24fps. But guess what? A telecine runs at 24/1.001 fps (23.98) so that it can transfer easily to NTSC. This slowing of
frame rate by 1/1000th is called PULLDOWN. Additionally a telecine adds a 3:2 field sequence to get to NTSC video. The combined process is known as a "3:2 Pulldown". This difference is so small that no one really would notice the difference if all we were looking at was the picture...
Now, lets take audio... for film it is usually captured in the field on an external recorder at 30FPS. Assuming perfect adjustments between all devices, if the film is run at 24FPS and the audio is running at 30fps, there would be no drift between audio and the image. But if the film has been pulled down, the audio must be pulled down as well (to 29.97fps). Audio houses and telecine facilities are used to dealing with pulldowns and pullups to keep audio in sync.
Now onto High Def... What if we want to integrate a 24P camera exactly into a film workflow (along with it's associated pullups and pulldowns)? Well, lets run the camera at 24 and edit at 24fps. Now we need to send an NTSC-VAPP (Video for Audio Post Production) to the audio house. We dub to NTSC by first performing a Pulldown (24 to 23.98) then add a 3:2 sequence and then downconvert it. This is sent to the audio house who takes the audio, performs a pulldown, and mixes to the NTSC-VAPP. Now when the project is finished, the audio house performs a pullup and sends it to be joined with the 24P master.
What if we don't like all these pullup/pulldowns... well we shoot at 23.98 and run our field recorder at 29.97. Then we edit at 23.98 and make the same VAPP for the audio house but with no pulldown (just 3:2 and downconversion). The audio house then mixes without pulling down the audio and delivers without pulling up the audio. WOW... no pullups or pulldowns though the entire process. (Now if we do perform a tape to film conversion, there will need to be a pullup of the audio at that time.)
From my observations the only reason to choose between 23.98 and 24 whether you need to fit seamlessly into a film workflow. I usually tell people that unless you are mixing film (other than for things that don't have sync sound or are shot
off-speed) and HD, then shoot 23.98 and leave all the pullups and pulldowns out of the equation.
Also, some rantings for the record
1. Panasonic's AG-DVX100 mini "a format that has nothing to do with
HD" camcorder does not shoot at 24, it shoots at 23.98.
2. Edit systems, as far as I have seen, are able to use 24P and 23.98 interchangeably, however you may have to pullup/pulldown your audio to match.
3. Pulldown is the process of slowing the frame rate. Many people use it to indicate that a 3:2 sequence has been added. They are different things, please use the proper terms.....
|As a stock footage company dealing with film, HD and SD we are constantly in the mode of delivering various tape and film formats to meet our clients post production requirements.
Recently I ran into a problem.
I sent a 720p HD tape shot at 30fps (Panasonic Varicam) to a post facility in LA and asked them to make a HDcam 1080 24 p conversion as per my clients request. Previously I sucessfully had converted footage shot at 1080 i (HDcam) to 1080 24p (HDcam) using the Panasonic UFC1800. My assumption was that since the Panasonic studio deck can seamlessly output a 1080i signal while playing back a 720p tape that in effect I was asking the post engineer to convert 1080i to 1080 24p. My logic was that since 30fps should make a mathmatically perfect 60 i and that it could be done inside the deck then this should work.
The post supervisor called me to say that he had tried to go direct from my 720 p 30fps through the UFC 1800 to HDcam 1080 24 p --he reported a stepping effect (the shot was from a moving vehicle) and attibuted that to the 2:2 frame sequence. So I suggested making the output of the Panasonic studio deck put out 1080i thus ,I thought, solving the problem. Why shouldn't this work?
|I guess that the short answer is that just running 30P through a box to "60i" doesn't make it 60i. Unless there are 60 discreet motion samples a second it will never be "true" 60i. Unless the process interpolates new motion samples, the 30P is just packaged in a 60i transmission (just like adding a 3:2 sequence doesn't make video 60i, it just packages 24P into a package that is transmitted at 60i). For 30P to 60i, each frame is used to create exactly 2 fields. The same is true of the 30 to 24 conversion but the math doesn't work out well. To do this conversion and keep the motion smooth, 75% of the frames would need to be interpolated frames. There are software packages available that attempt these interpolations with varying success. With a limited number of motion samples available, it becomes very difficult to keep the motion smooth as well as the spatial resolution high (not just blurring frames together).
The cleanest looking solution would be to just slow the 30p to 24p for a result of 80% of real time slowmo (if that is an option for this particular footage).
This topic could probably fill a book...