Is it acceptable for syncing be off by 1-2ms in a 24P Music Video?

tbagg

Member
Hi -

[Short Question]
I am editing a music video and I've filmed several 4min takes of someone rapping at different angles using my DVX100b. After lining them up in my editing program as close as possible to the master track I zoom in and they are off by 1-2ms. Is this acceptable? or is this amateur?


[Long Question]
I lay down the mastered audio track – and then I have 4 video tracks, each video track has a full 4 minutes of the guy rapping. I line up each of those to the master as close as possible. Then I put a BLACK MATTE over them and start my actual video above the matte layer. ie. I can razor the footage from below (video 1-4) and using the magnet tool and bring it up to VIDEO 6 so it will line up where it’s supposed to.

Here is a diagram of my editing timeline:

VIDEO 6 – MUSIC VIDEO LAYER
VIDEO 5 – BLACK MATTE BACKGROUND – stretches 4mins of timeline.
VIDEO 4 – rap 4 – stretches 4mins of timeline.
VIDEO 3 – rap 3 – stretches 4mins of timeline.
VIDEO 2 – rap 2 – stretches 4mins of timeline.
VIDEO 1 – rap 1 – stretches 4mins of timeline.
--
AUDIO 1 – rap 1 wav peaks (music cranked from car) - audio disabled
AUDIO 2 – rap 2 wav peaks (music cranked from car) - audio disabled
AUDIO 3 – rap 3 wav peaks (music cranked from car) - audio disabled
AUDIO 4 – rap 4 wav peaks (music cranked from car) - audio disabled
AUDIO 5 –
AUDIO 6 – MASTERED CD AUDIO (this is lined up with the wav peaks from audio 1,2,3 and 4 – as close as I can get them that is)


I use the wav peaks from audio 1-4 to line up the video tracks as close as I can. They aren't perfect because here is what happens when I zoom in.

/\ = First Beat of song.

[Frame 1 ][Frame 2 ]

[Audio 6 – Master ] /\
[Audio 1 ] /\
[Audio 2 ] /\
[Audio 3 ] /\
[Audio 4 ] /\


Note that audio 2 is ahead by 1ms about – which means that video 2 is ahead by 1ms, audio 3 is behind by 1ms so that means that video 3 is behind by 1ms roughly etc.

My question:

The only way to avoid this is during recording – somehow I would have to make the dvx100b always record the first beat 1ms into the first frame frame. I don’t know if that’s possible with a DVX100b or only expensive equipment. Also - I don’t know if my friend and I are being too picky? maybe it is normal for video to be slightly off? Maybe 1ms isn’t noticeable but just because we have zoomed in and we see that things aren't lined up we find it noticeable.

So – basically – I guess my questions are – does anyone know how I can re-record and sync it up perfect? OR - are professional videos always off by a few ms when you zoom right in?

-Trevor

PS - the video was shot in 24P.
 
Ummm... I have no idea what's going on here (probably because of the massive hangover I have), but the short answer is... NO, everything should be PERFECTLY in sync. Your speaking in miliseconds (I think), but you should be speaking in frames, as you can always adjust your footage one frame at a time, getting you perfectly in sync.

Your post really confused me at the moment, but I think I may get the jist of it.

First off... what did you use to capture the footage? the DVX itself? Lots of times syncing issue's arise during the capture process.

Secondly, delete all the audio *except* the master audio track. Third - - snyc the footage.

Easier said than done, huh? not really... find a point where something lasts only one frame (a clapboard would have been nice), like a camera flash or something, and sync all the footage to it. With a little practice, you can do this by eye (although I don't recommend it unless it's necessary). Just watch the singer and move your video around ONE FRAME AT A TIME untill it's perfectly synced.

also, what NLE (editing software) are you working with?

EDIT: I just read through your post again and I'm still a little confused about your editing process. Try this - - sync up all your video tracks (to the frame) on top of eachother (sounds like you're close) > delete your black matte (as I have no idea why it's there in the first place) > now you should have 4 video tracks just sitting on top of eachother (in sync), and one audio track, correct? now instead of finding your shot and using your "magnet" to pull it up to a seperate track, just cut everything else out. You should be cutting bits and pieces out of each track untill you have your rough cut. That's your first edit.

Now that you have that, you go back to the beginning and start over (I usually start a new sequnce and copy and paste my *rough* cut into it, leaving your first sequnce as is). Now you begin your secong cut. Watch it all the way through first to get an idea of what needs to be fixed, and start making changes. Repeat this process untill you're completely satisfied with the edit (making NEW sequence's along the way so you can always go back to a previous cut if needed).

Then take your final edit, start a new sequence, and polish it up. Add color grading, effects, transitions, etc. Repeat as necessary *rinse and repeat*

Don't know why I just rambled this entire post, but hopefully you'll be able to get somethig usefull out of it.

-brown
 
Last edited:
Thats the one thing people notice right away, especially with music videos. I would say its amateurish. Why put yourself out there with something that might tell other people that you don't know what your doing? Didnt you get a scratch track of the song playing when they performed? I think it would be easy to hide though since music videos can have a ton of quick cuts and maybe have each cut = a bar of music.
 
rcbrown23 said:
Your speaking in miliseconds (I think), but you should be speaking in frames, as you can always adjust your footage one frame at a time, getting you perfectly in sync.

Subframe accuracy might be necessary, I think, if tbagg is trying to sync audio from one take to video from another, but it is not clear to me what tbagg is trying to do and why he has numerous audio tracks. Tbagg, if you could describe what you are trying to with each audio and video track and what NLE you are using, someone here who shoots music videos may be able to help you.
 
Hi Guys - I just realized I lost my spaces - AUDIO 1, 2,3 and 4 weren't supposed to line up perfectly. Here is how the diagram was supposed to look:

23.97 frames per second / 100ms per second = approx 4ms a frame.

/\ = First Beat in the song.

<--------------- 4ms ------------------------><--------------- 4ms ---------------------->
[Frame 1_______________________][Frame 2______________________]

[Audio 6 – Master ]_______/\
[Audio 1 ]__________/\
[Audio 2 ]_/\
[Audio 3 ]______________/\
[Audio 4 ]______/\



a.) Audio 1-4 are only on the timeline to help me sync it to the master. I don't actually have it playing.

b.) I captured all my footage with the DVX100B - The MASTER is a CD Master supplied by the artist.

c.) The artist is my friend so luckily I am not pissing off a client. We can re-shoot no problem. I am doing this for free so he won't be mad.

d.) I am using Premiere Pro 2.0


Is it possible to recapture so it looks like this:

<--------------- 4ms ------------------------><--------------- 4ms ---------------------->
[Frame 1_______________________][Frame 2______________________]

[Audio 6 – Master ]_______/\
[Audio 1 ]__________/\
[Audio 2 ]__________/\
[Audio 3 ]__________/\
[Audio 4 ]__________/\

Somehow when I hit the record button on the DVX100B - the song would have to start at the exact same time from when I hit the record button to make this happen. That way it would be accurate to the ms. within the frame.

Thanks so much for the help so far. rcbrown23 - i appreciate the rinse/repeat tip and I will do that for sure.
 
tbagg,

Sub-frame accuracy usually isn't necessary for music videos. Beyond that, there is basically nothing you can do once its recorded to match that up.

Slightly different point now, you can't necessarily trust the raw audio from the footage you shot because the audio source was in the car, and the talent was in a different position, and the camera was somewhere else. All you are doing is lining up the audio tracks, but the lip sync itself might not be spot on because of those circumstances.

So, you have a good starting point in lining up the wav's. However, now you should probably go watch each video track with the master audio from the cd, and see if it actually matches up. You may still have to nudge the video a frame or two in either direction to actually make the lip sync match. The wav may end up a little off, but the location audio isn't what matters, it's the lips that count.

Josh
 
The golden rule in syncing footage is NEVER let your sound drift AHEAD of your picture - it's a dead giveaway. If it drifts behind the picture, you can often get away with 1 or even 2 frames. In the days of the Moviola, when we didn't have subframe accuracy, if we had audio that was drifting back and forth (say it was shot with an ungoverned camera), we would sometimes knock the entire track back a frame just to be safe.
 
You can use only one audio track (master CD). You can only match the video tracks with 1/2 frame accuracy (and rather have sound ahead than behind). Live with that, NOBODY will notice ANYTHING.

You can of course use the extra audio tracks to align the video, but no more.
 
Now we're hearing two contradicting posts from cyclone and petrus. Petrus says if it's off a frame or two, he'd rather it be ahead - - - cyclone says it should be behind

??

Which is correct?

I'd have to guess with cyclone, saying behind is better... but it's just a guess based on what I've seen in my own video's.
 
I can only speak from my experience, but just last week I was cutting a spot and someone said, "Ya know, the girl speaking just doesn't seem quite in sync to me." so we knocked the whole track back 1 frame and the concensus was that the spot now looked perfect. As I mentioned, I've seen this happen quite often; while I've never seen advancing the sound solve the problem. My guess is that since light travels faster than sound (299,792,458 miles/sec vs. 340 miles/second), in real life we never hear the sound of an action first, even by a fraction of a second. But we often hear the sound quite a bit after the action (speaking, gunshot, lightning and thunder, etc.), depending on how far away we are. So no matter how close or far away we are, since we have never heard the sound first, our brain just doesn't accept it, even by a few milliseconds, when we hear it early. Of course, when people are speaking, it is only apparent on hard consonants like "p's" and "b's" and "m's".
 
Those units should be METERS per second, not miles per second..sound travels about 740 miles per hour and light travels 186,282 miles per second. I think your explanation of how our brain accepts delayed audio better than premature audio is right on. I'm no pro in the music video business, but it seems to me that even if a half-frame discrepancy is present ( 20 milliseconds in 24P) if it's noticeable you can pad or trim the audio by that much in an audio editor and re-import it into your video editor. What may be more troublesome is if the overall tempo of the video is different than the CD track...which is why the CD track has to be used as the tempo guide for the video performance. If it's not, the synch cannot be exactly on for more than a few frames. Multiple edits in the video and independent alignment of the segments will keep it as close as possible if that's what you want.
 
COOL!

Thanks guys for helping to push forward my 'stalled' project. I have appreciated all of your comments like the light/sound, the 20ms half frame fact and how I learned the new term 'subframe'. To sum up the thread:

Q: "Is it acceptable for syncing to be off by 1-2ms in a 24P Music Video"
A: "[Yes,] the wav may end up a little off, but the location [of the] audio isn't what matters, it's the lips that count"


A few more things I am wondering about:

a.) Do any of you have a link to a 24P music video (that doesn't use sub frame accuracy)?

b.) To help push forward my camera skills - how would I get sub frame accuracy if I ever needed it some day? (although I have no idea when I would need it)

c.) Can I assume that the DVX is a 'governed camera' and that 'audio drift' and those types of problems are related to older cameras? ie. Issues people had 10+ years ago that have since been resolved.

Thanks again everyone - you don't know how helpful you've been!
 
Good point about meters vs miles, bherr. I knew I was pushing my luck when I tried to get fancy and take a stab at the speed of sound and light.
 
My mistake, writing late at night... Sound can lag behind a bit, never place it ahead of picture, unnatural.

My main point was that one frame accuracy is the best you can get unless you have cameras synchronized. In normal productions they never are, as one frame misalignment is no big deal especially in a fast paced music video.
 
Back
Top