footage drifting from Zoom H4N

bill totolo

bill totolo

Veteran
Shot night Ranger performing at the NAMM show with my 5DII and recorded audio with a Zoom H4N.

Imported footage with Lexar FW800 CF reader, stored on internal Seagate 7200 drive, transcoded into ProRes422 LT using MPEG streamclip and drag and dropped into FCP7 bin and onto a 30P timeline.

Audio converted to QT movie w/ QT 7.6.3

Line up audio wave forms a the beginning of the clip but they drift about a minute into performance and progressively gets worse.

Thoughts?
 
You need to conform the video to 29.97 (cinema tools takes 2 sec) and work on a 29.97 timeline. Then you should be golden.
 
i have read this on philipbloom blog today:

Problem is sound drifted after about 10 seconds and I couldn’t for the life of me work out why. Both 7D sound and sound on Zoom (24 bit WAV) were at 48kHZ so there was no reason for any drift to happen. The sequence settings were correct, they matched the video perfectly so WHY was it drifting? This freaked me out for 2 days…I have never had this happen to me before. Solution came from Robin Charters, son of Rodney Charters who said as I had been in the US did I perhaps at some point have a 23.98p set up in my FCP…the answer was yes. He said that I needed to make sure my capture preset was also the same as the sequence settings as Final cut pro has a glitch that insists that the capture preset be set to same frame rate. So my frame rate is 25p therefore my capture preset needs to be 25p too. Not necessarily 25p Pro Res LT 1920×1080p as my footage is, even PAL 25p would do as FCP imports audio into sequence from this setting NOT your sequence settings. This has been plaguing people for months, Bruce Sharpe has always recommended using easy set up but this never explained what was causing the problem…now thanks to Robin we know why. Robin will be writing an in depth article about this soon and I will share the link with you as soon as he has done it.

With this VERY handy bit of information I redid Pluraleyes and low and behold we have perfect sync. Zero drift and all the interviews are synced up (apart from one bit where I forgot to turn the RODE mic on so we had no reference audio, schoolboy error that one!)

I hope this was useful to you as there will be many more supertips and hints like this to come![/I][/I]
 
bill totolo said:
Imported footage with Lexar FW800 CF reader, stored on internal Seagate 7200 drive, transcoded into ProRes422 LT using MPEG streamclip and drag and dropped into FCP7 bin and onto a 30P timeline.
Click to expand...
You said 30p, but is it 29.97 or 30.00? The standard for video is 29.97, but the 5D doesn't shoot 29.97, it shoots 30.00. So if you're editing at 30.00 you shouldn't have any drift, but if you're running your 5D footage at 29.97 instead, then yes, it'll drift.
 
That's interesting Shiny4. I think that applies more to European 25p land more than it does here, because there's not much of a reason to work at 30.000 fps when you'll eventually need to conform to 29.97 for dvd's, broadcast... etc. Good to know nonetheless.
 
Barry_Green said:
You said 30p, but is it 29.97 or 30.00? The standard for video is 29.97, but the 5D doesn't shoot 29.97, it shoots 30.00. So if you're editing at 30.00 you shouldn't have any drift, but if you're running your 5D footage at 29.97 instead, then yes, it'll drift.
Click to expand...


Maybe I'm wrong and what PBloom discovered is really what's going on under the hood in fcp, but I think the H4n actually records audio at an equivalent speed* of 29.97fps and not the speed of 30.000p. Notice I'm not talking about framerate. It doesn't have a framerate. I believe it does record at a Speed, and they made it so it will sync with NTSC devices, which is smart, I think. The PBloom workaround will make it sync with the non-slowed video of Euro-land.

*29.97 is really 30.000fps SLOWED down to run at 29.97 fps, hence the need for drop-frame timecode to make up the difference between 1 real second and 30 frames of video. (not the same length) There's no difference in Speed between 29.97 and 23.98, only a difference in the number of frames in that long second.

I'm sure someone will correct me if I'm wrong here, but I wish we could have a syncing sticky that would put this issue to bed.
 
What Philip was talking about was a 7D, not a 5D. The 7D shoots at 29.97, the 5D shoots at 30.00. That's a big huge fundamental difference.

Audio isn't recorded at any framerate, you can use the exact same audio file with 24p footage, with 30p footage, with 60p footage, it doesn't matter, so long as your NLE and your audio file are set to the same sampling rate. Just like your footage and your NLE have to be at the same frame rate (29.97 or 30.00) your NLE and your audio have to be at the same sampling rate (i.e., 48000 Hz).

Drop-Frame timecode doesn't change anything about the way the video works or how it runs. It's simply a numbering system that drops a few numbers here and there to make up the 0.1% difference between 30.000 and 29.970.
 
Yes, but he was shooting at 25p, no? That's the same equivalent speed as 30.000p or 24.000p 60.000p

Then, when he imported the audio with his capture settings at 23.98 the audio wouldn't sync. BUT, when the capture settings were at 25p, the audio would sync.

SO- It must be conforming the audio to 25p on import and 'giving' it a framerate, (really it's conforming it to a speed that syncs with all XX.000 framerates) which implies that it had a framerate/ speed to begin with, (which is a speed that will sync with any XX.97/ 98/ 95 framerate)

All of my experience with the D90, GH1, and 7D syncing with the H4n has shown me that it records at the slowed rate equivalent to 23.98, 29.97, and 59.95 and will sync with anything shot at those rates. But then again, it could just be that my capture settings have always been on one of those rates so it's importing that way because of that.
 
If picture is at 30.000fps then record audio at 48.048kHz - when working at the NTSC rate of 29.97, then both with be slowed down .1% and remain in sync where 29.97 will be 48kHz. Both picture and sound need to be adjusted - of only picture it will drift over time.

Michael
 
Tried re-conforming to 29.97 from 30p, no difference. I know I'm missing a step.

If someone is doing this successfully with the 5D and the H4N can you give me all the A/V settings so I can go thru them one at a time?
Sorry for being a pain, would love to resolve this for future projects.

Thanks,
 
Last edited:
OK, I have a workaround.

Instead of opening the wav file with QT7 and saving as a self contained QT movie I simply dragged the wav file into the sequence. Everything is now in perfect synch.

I thought in order to edit audio files I had to save as QT movie files. Maybe that's just for MP3's?
 
You must log in or register to reply here.
Back
Top