Hi!
I work with TV series every day, and every day I have to import AAF with 1024 samples offset (this is a feature of MPEG-4 v.2 codec used by the editor to upload the video for me)
So, every time I import an AAF, I have to put the cursor 1024 samples ahead to make the sound from the AAF match the raw sound from the video.
regards, Alex
Sorry for being an absolute tool, but - is it the video audio that has correct sync and not the AAF audio (ultimately)?
In other words, if you shift the AAF and send back a .wav file to the client and they lay that in and export to a different master format, will sync hold?
Short explanation:
This offset occurs when audio+video (assuming it is uncompressed) is encoded in lossy format.
If I don’t compensate this delay when importing AAF (and ignore the fact that the sound in the video doesn’t match the sound in AAF) then in the end, the editor will get 2x offset (2048 samples) if he imports my sound and renders the video in the same lossy codec.
This problem is relevant only when lossy codecs are used. But considering the volume of production it is a reasonable choice for sharing and storing material
regards, Alex
I would have thought that if you left AAF as-is and then exported the mix it would lay back the same as the source audio in the sequence. In other words the mix would match the audio on the timeline in the sequence that generated the AAF. Therefore the offset would be the same, not 2x.
I guess my question about this more broadly is about what responsibility we have in audio and how we solve problems. If for example they have a master sequence in which everything is in sync and then render out different masters and videos using different codes then you have zero control over the difference between those, only they have control. So your adjusted -1024 audio would possibly line up correctly only with this one target codec and would be off on others. And if some are off is it reasonable to ask you to adjust the mix for various codecs?
I understand of course that if they have specified that this is the video we need sync for and this is what we’re outputting then it is what it is and you do the job, so I’m not disputing that. Just thinking about this more broadly.
And I +1 your request I think.
For now you can maybe make a macro to quickly move the cursor?
I just came up with this idea - just put a marker (in a template) at the desired offset, and in the Macro go to that marker and call the import of the video.
BTW protools has an offset option in AAF import
regards, Alex
I agree with Mattias, as someone who works on both sides of post (I am primarily on the video side, actually.)
The offset is coming from the MP4 which itself is wrong (it’s the decoder delay with compressed codecs like AAC) but you’re moving your audio to match a bad file. So you send back a mixdown WAV which gets put back on the timeline the original AAF (which shouldn’t be screwed up) came from… and if they export a new MP4+AAC from that, it will have the same delay the one they sent you originally did, and they should match each other.
A lot of players are aware of the decoder delay and will make up for it on playback, but in this case Nuendo doesn’t when you import the audio from the MP4, so you get this as silence at the beginning. But compared to the picture, it shouldn’t actually be there… so you’re shifting your AAF audio to match a bad audio reference track.
Here is the beginning of a commercial exported from the same original timeline in my editor as ProRes with uncompressed audio, and as MP4 with AAC audio, then both resulting movies imported into Nuendo extracting the audio:
The ProRes/PCM audio is correct. The AAF import isn’t shown but it matches that, and if I export it from Nuendo as WAV and put it back on my video timeline and invert the phase, they cancel each other out exactly. The MP4/AAC audio is delayed, yes. But it doesn’t represent reality… If I “fix” the AAF to match that, export a mix, put it back in my video editor,and export a ProRes for final delivery, I’ve now added a delay to the audio as it started out life in the timeline the AAF came from.
I think you’re fixing something that doesn’t need fixing.
But all that said… 1024 samples is nothing either way 
