NTSC and PAL

[pboong]

Hi guys,

Comparing with NTSC dp and PAL, which seem to have a longer duration?

Because in this case, I have an audio format given by client whereby it does not tally against the video. Which, we highly suspect that the audio follow the NTSC timing whereas the video is in PAL. (I had captured the video using PAL DigiBeta)

Given that, there is a 5min difference in this 2hrs show.

NTSC Runtime is 124.35 and PAL Runtime is 119.29.

I used the WFT Timecode calculator to make a simple calculation to check the timecode and it seem to tally, the audio source given is NTSC timing rather than PAL.

As such, what is the proper method I can use to convert the audio timing to match up with NTSC duration? Does Protools HD 7 allow me to change the audio timing?

Appreciate and thanks!

[Calidore]

Hi, pboong. PAL video is 25 frames per second, and NTSC is 24 fps, so PAL videos run 4% faster. Exactly what format are your audio (MP3/AC3/WAV/etc.?) and video (MPG/MP4/AVI/MKV/etc.?) files in?

Best,

Calidore

[pboong]

Ain't NTSC is 29.97?

If PAL is faster than NTSC by 4%, it means the audio duration for NTSC is longer, right?

I suspect my audio follow NTSC video as i gotten from my client.
On DigiBeta tape, it is in PAL.

When I align the audio and video together, the audio is longer than the video.

Hence, suspecting the audio follow the NTSC. My client are unable to give me the audio in PAL tho. So, I am asking if by using Protools time shift, will it works if I set the End Timecode properly?

Or using Soundforge is better?

[Alex_ander]

Ain't NTSC is 29.97?

Of course, it is. Talking about 4% difference in length, people mean conversion between PAL/NTSC for film source which is 24fps (modified to 23.976 with pulldown to 29.97 for NTSC). To keep frames unmodified at conversion to PAL, first 23.976 is restored, then it is sped-up to 25. This conversion method makes PAL 4% shorter.

[pboong]

Ain't NTSC is 29.97?

Of course, it is. Talking about 4% difference in length, people mean conversion between PAL/NTSC for film source which is 24fps (modified to 23.976 with pulldown to 29.97 for NTSC). To keep frames unmodified at conversion to PAL, first 23.976 is restored, then it is sped-up to 25. This conversion method makes PAL 4% shorter.

Oic. Thanks Alex, now I get your meaning tho.

For instance, right now, I have the digi beta tape (PAL) been captured to my PC using Blackmagic.
As such, I have 2 set of audio to match with the video too. The first audio, which is 5.1, is 100% match with PAL video. The 2nd audio, which is the commentary, is matched with NTSC video. Hence the difference show in timing.

Now I am using Time Shift in Protools, I set the end timecode to match up with the video. It is slightly drift as I notice the percentage is 104.+%.
In what way, can I really match the audio perfectly along with the video?

Initially, this audio is given separately with the video tho.

[manono]

It depends on if the commentary track came from a drop-frame source (like a 23.976fps DVD) or a non-drop-frame source (24fps). If drop-frame, then 25/23.976=104.271%. If non-drop-frame then 25/24=104.167%. My guess is it's drop-frame, but you could try them both to see which matches up better. One should be exact while the wrong one will be off by about 7.5 seconds by the end of the movie.

[pboong]

This is quite strange on my end.

On Protools, I had set the percentage to 104.2% but it still does not sync with the video. My video is 25fps, whereas the audio at NTSC dp.

I was wondering if this happen works.

Having my current video in PAL, I convert to NTSC using TMPGen Express 4. Then I shall check if the audio and video sync.
If it does, I will drop in again to that program and do the conversion of NTSC to PAL and set to export audio only.

Will that works?

I just had a trial earlier on. I have no idea why is it after converting, it is still at 25PAL rather than NTSC.


Can I actually import the video to FCP or Adobe Premiere Pro, render out at NTSC?

[edDV]

Leave video as is, just work on audio speed conversion. Use Soundforge.

From the program length calculations, it appears the original camera shoot was 24p (or 23.976p). Your PAL video has been increased in speed 4% (25/24 = 1.042, 25/23.976 = 1.043). You now need to speed up the audio to match. Soundforge will change audio speed without changing pitch. This is all standard procedure with film to PAL.

I'm assuming from above your goal is a PAL tape. Why are you rendering to NTSC? The only NTSC issue is whether the camera was shooting 24p or 23.976p.

[poisondeathray]

If the original shoot was 24p , and you captured as 25p , you will have repeated frames or blended frames in the video.

I wouldn't be a happy client.

Why not recapture properly , then do a proper NTSC=>PAL conversion if that was your goal

[lordsmurf]

I suggest Goldwave for time-based stretching, as you can do it with SMTPE clock (hour:min:frames) ...

... instead of the silly % basis found in SoundForge.

[edDV]

I only mentioed Soundforge because he said he had it.

Most pro shoots will use a time code slate to help sync audio. Audio is recorded separately but with the same time code on a track. That way everything can be matched in post to the frame.

[davexnet]

On my version of Sound Forge (version 8) you can enter either a percentage or absolute time.
There are a bunch of different algorithms to choose - they all seem to degrade the sound, unfortunately.

[edDV]

What format is the sound?

[davexnet]

Pcm/wav.

[pboong]

Thanks all for the reply. Perhaps let me explain in details.

First, my job is to capture the feature from digi beta tape. It is already in PAL.
As I have to send the feature in mpeg1 to my client for Chinese subtitling, I had it captured using optibase.

For .mpeg, I had the LTC included.

There are 2 audio involved in this scenario. 1, the audio from the feature in digi beta.
2.Its commentary audio sent by client.

As the fact, I tally against the commentary audio to video, my audio is longer than the feature. Which I am sure that the audio is tally with NTSC.

Now, what am I suppose to do is to get the audio match with the video.

The reason why I want to get the video in NTSC is to get the audio match up for confirming.
I thought by matching up the NTSC audio and video , I can convert it to PAL using TMGenc 4 xpress. All I want is audio.
For video, I am using the original one. But I have to tally to see if it match up.

I myself have no idea if it will works , thus seeking help.

If I use soundforge, do I increase the audio by 104%?

Is goldwave a better program to use?

Pls advise tho...

Really appreciate a lot tho.

[edDV]

If the original master was 24p (e.g. most theatrical films) then speed up 25/24 = 1.041667

If the original was mastered as a TV series in North America it was probably shot and mastered 23.976*. In that case you would speed sound up 25/23.976 = 1.04270937 to match PAL 25 fps video.

Fortunately, you don't need to worry about lip sync with a commentary track.

Soundforge should do OK from PCM source. I haven't used Goldwave.


* If you received the audio on NTSC DigiBeta, the audio duration would be the same for telecined 23.976 or 29.97 fps source. It is true that drop frame vs. non drop frame timecode mode can make a 3.59 sec difference per hour in the way timecode duration is measured but the video or audio duration as measured on a stop watch is the same.

[pboong]

Hi edDV,

Apparently, the feature I gotten is a movie, "Iron Man 2".

And, the audio in NTSC is not from Digibeta either. It is the soft copy given by my client.

Although I know that I do not need to worry so much about the lip sync on commentary, but on the background of the commentary involve the actual show sound. Hence, the adjustment must be accurate tho.

If I were to use Soundforge to adjust the audio, I have to use Time Stretch?
Since film to 25fps is about speed of 1.041667, I set the percentage of 104%?

Edit:

Right now I'm using Goldwave. Do I use time warp and set the percentage to 104.1667%?

Add on:
After doing that, it seem like the audio is still slightly drifted. Video and audio started at 01:00:00:00. Video (PAL) End Timecode: 01:59:28:01 / Audio (Speed up using Goldwave at 104.1667%): 02:59:45:24

In fact, I just realised after playing the last few second of the video, I guess that it drifted by 3-4 seconds throughout.

[Calidore]

The precision has to be fine enough that the sync holds up to the end. When I'm converting PAL to NTSC with Virtualdub, I have the audio stretch ratio set at 1.04271.

Best,

Calidore

[pboong]

The precision has to be fine enough that the sync holds up to the end. When I'm converting PAL to NTSC with Virtualdub, I have the audio stretch ratio set at 1.04271.

Best,

Calidore

Could you tell me how you use virtual dub to do the conversion? Right I wanted NTSC audio to sync up with PAL video.

[edDV]

The precision has to be fine enough that the sync holds up to the end. When I'm converting PAL to NTSC with Virtualdub, I have the audio stretch ratio set at 1.04271.

Best,

Calidore

Could you tell me how you use virtual dub to do the conversion? Right I wanted NTSC audio to sync up with PAL video.

You're just going to have to adjust the speed for that 3.5 sec per hour error. We can only speculate how the audio error happened. Trial and error time.

[pboong]

Currently, by using Goldwave, it can show me the end time code when I adjust the percentage. However, the downside is I have no reference from NTSC video that show when the audio end. So, right now, I can only make use of trial and error from the percentage of 4.1 to 4.4%. hoping I can get it down asap.

Hoping it will not take me too long to get it solve tho.

[edDV]

Currently, by using Goldwave, it can show me the end time code when I adjust the percentage. However, the downside is I have no reference from NTSC video that show when the audio end. So, right now, I can only make use of trial and error from the percentage of 4.1 to 4.4%. hoping I can get it down asap.

Hoping it will not take me too long to get it solve tho.

Two ways to go next, analysis or trial and error.

On the analysis side, 3.5 sec over 1 hour is the difference between NTSC drop frame and non-drop frame time code. Instead of time code, you can use a stop watch.

Or, just zero in with trial and error adjustment. Audio processing is fast.

Note the final settings in case the client wants to change something. They always will.

[pboong]

@edDV - I had finally solved it. I had use the 1.04270937 instead tho. Everything is fine now.

As you mentioned earlier on, filming is mastered at 24p and tv series mastered at 23.976p. How come for this case, Iron Man 2, is in 23.976? I thought 23.976 is usually the framerate they filmed it?

[edDV]

@edDV - I had finally solved it. I had use the 1.04270937 instead tho. Everything is fine now.

As you mentioned earlier on, filming is mastered at 24p and tv series mastered at 23.976p. How come for this case, Iron Man 2, is in 23.976? I thought 23.976 is usually the framerate they filmed it?

It could be either by the time it reaches you.

[Cornucopia]

If you've got PT HD 7, the time stretch in it is better than anything you're going to get with any of those other apps - and I've used them all thoroughly.
With PT, you can do it mathematically, but you could also have done it graphically by lining up the front end with time ZERO and using the trim tool (in timestretch mode) on the back end to line it up EXACTLY to the frame of video (which you should have had imported into the video reference track).
IF after that, the audio has sync inconsistencies IN THE MIDDLE of the piece, it'll probably NEVER get completely synced up, because that usually means there was a problem with the audio recorder's internal crystal clock.

This is a moot point as I see you got it fixed, but maybe when the next job rolls around...

Scott

[pboong]

Thanks all!

Cornucopia - I thought of doing that too. I didn't know Protools allow me to do the real time timestretch tho. I do know Cubase able to.
I just give that a try for trial.

BTW, what happen when I wanted to convert the Video from NTSC to PAL without losing much of quality.

[jagabo]

@edDV - I had finally solved it. I had use the 1.04270937 instead tho. Everything is fine now.

25 / 23.976 ~= 1.04270937

The slightly more accurate adjustment would be:

25 / (24000/1001) ~= 1.04270833

The difference would only be about 7 milliseconds over a two hour movie.

As you mentioned earlier on, filming is mastered at 24p and tv series mastered at 23.976p. How come for this case, Iron Man 2, is in 23.976? I thought 23.976 is usually the framerate they filmed it?

Film is shot at 24 fps. It is slowed down to 24000/1001 (~23.976) fps when telecined.

[pboong]

actually, what is the exactly telecined?
And why they wanted to slow down to 23.976 instead of 24?

something i would like to know more of.

[jagabo]

For technical reasons having to do with the way color was added to the original black-and-white TV signal, NTSC color TV is (was) broadcast at 60000/1001 (~59.94) fields per second, not 60 fields per second. To produce 60000/1001 fields per second from film it is slowed down from 24 fps to 24000/1001 (~23.976) fps then the film frames are displayed for either 2 or 3 fields, alternating between the two. Ie, the first film frame is displayed for 2 fields (~2/60ths second) then next is displayed for 3 fields (~3/60ths second), then 2 fields, then 3 fields, etc. On average, each film frame is displayed for 2.5 fields -- 60 / 2.5 = 24, or (60000/1001) / 2.5 = (24000/1001), or ~59.94 / 2.5 = ~23.976.

[edDV]

actually, what is the exactly telecined?
And why they wanted to slow down to 23.976 instead of 24?

something i would like to know more of.

Without sinking into too much detail unless you want to, I'll give the business reason. The business goal is to shoot and edit in a single format that can be directly converted to all the common world film and TV formats and be compatible with legacy format archives.

For historical technical reasons the world has a common 24 fps film format and different regional television formats. In the analog days, NTSC and PAL transfers from film required a completely separate transfer and edit because of resolution and frame rate differences. HDTV standards have consolidated world TV resolution standards to 1920x1080 and 1280x720 leaving only frame rate differences between the formerly NTSC and PAL/SECAM regions.

As Jagabo explains NTSC television uses 30/1001=29.97 as a frame rate base. This determines the following versions of NTSC derived HD formats used today.

23.976 fps (29.97*24/30) for progressive film rate
29.97 fps for interlace broadcast
59.94 fps for progressive broadcast

For historical reasons, the PAL/SECAM regions use a 25 fps base for interlace and 50 fps for progressive.

Here is a key point... There is no direct conversion possible between 25fps to 29.97fps or 25fps to 24 fps.

In other words 25 fps is regionally isolated and can't be easily converted into anything else. Same goes for 29.97.

So we are left with 24p as the legacy base to build a world distribution empire. Whatever artistic defense you have heard for 24p, it is all rationalization. 24p was originally chosen is still used for business reasons. You may want to argue techincal or artistic reasons to do otherwise but those arguements didn't fly in the HDTV or Digital Cinema standardization process.

So if you are going to shoot "24p" for worldwide film and television distribution, how would you define a digital master frame rate? You would choose the frame rate that directly converts as much as possible. That frame rate would be 23.976. You would transfer legacy film at slightly slow 24000/1001 ~=23.976 and you would use 23.976 for all electronic production and editing. The finished master could be converted to all the current and lagacy world formats as follows.

NTSC region:

23.976 to 480i/1080i 29.97 fps using pulldown (aka "telecine")
23.976 to 480p/720p/1080p 59.94 fps using a 3:2 frame repeats

PAL/SECAM regions:

23.976 to 576i/1080i 25 fps by frame interpolation (speedup)
23.976 to 576p/720p/1080p 50 fps by frame interpolation (speedup)

Digital Cinema: No conversion necessary

Legacy analog film conversion: 23.976p would run 0.001 % fast on a 24 fps projector

An added business bonus is shooting digital 23.976p doesn't upset existing unionized craft relationships or require extensive retraining. It is just like shooting 24p film.