OK...that sounds like a plan. Thank you for the advice.

I wish SONAR would get a 23.976 timecode option.

Technically, as long as the video playback rate is accurate, I can compose the score to it with SONAR and pray that the exported soundtrack lines up when imported back into Vegas. [A full-length project is the only way to really test the sync.]

So, I suppose I CAN work with the ruler and frames misaligned; but the misalignment makes me question whether I can reliably compose at a frame-accurate level.

There's a simple enough way to do this in Vegas. I've posted this before:
Switch T/L to Absolute Frames. Note down the number of frames at the end. Switch to 24p. Now stretch vision and sound to the exact same number of frames. No resampling involved but you might want to switch project audio resampling to best. Using values such as 0.001 you may have problems due to rounding errors.

If you render that out it'll be lossless, no interpolation etc going on and you can drop that onto a 24p timeline or into Soanar I guess and everything will line up to perfection. When everything is locked and finalised you can come back to 23.976 the same way if needed.

If you're working with mpeg-2 then you might suffer going down one generation however all you should need in Soanar is a vision proxy that gets ditched once you're done composing.

One way to avoid any real concerns is to always render full tracks of audio or better still polyphonic wave files that Vegas now handles very nicely. Everything is locked together, if you need to adjust differences in frame rates it's very simple with zero risk of sync being lost.

Bob.

Thanks Bob!

Given the situation, would you prefer to follow the procedure you suggested, or just import the 23.976 + audio into SONAR and ignore SONAR's ruler & thumbnails?

I think that Bob's approach will only work with certain types of video that Vegas can "smart render." This includes DV and, sometimes with V8, MPEG-2. I think his approach does the same thing as what I suggested (simply changing the FPS header info). My approach takes about 0.1 second for each video file. I believe that his approach will take quite a bit of time, depending on the speed of your hard drives. I would budget between 15-30 minutes for each hour of video.

However, try both on a small video segment and see what happens, then report back.

What happens to the video really doesn't matter, you should only need it for reference in the audio application. A lot of audio post houses still use SP and I doubt they strike a 35mm print from that SP dub!

The audio on the other hand does matter!

What I'd do is using a Vegas project retime it as I mentioned. Render using whatever video codec works in Soanar, just Preview qulaity might be good enough, it is only a reference file to carry the TC as Vegas doesn't do TC in audio.

With this file(s) in Soanar everyhting will line up, TC, the works.

When you do your music in Soanar render it out from there, a polyphonic wave file would be good if you want to mix in Vegas.
Only thing to note is your file from Soanar will be for 24fps.
You bring that back into Vegas and reverse the procedure.
Your camera tapes are now not resampled, and your original vision is back at 23.976, no harm done.
You will have to resample your audio out of Soanar no matter what you do unless the small errors you'll get over an long segment don't matter and they might well not, you'll have to judge this based on what you're doing. If you need frame accurate over 1 hour I think you'll need to. If it's just 5 minutes of audio/music to lay back into Vegas I can't see an issue. Many Hollywood movies seem to have music tracks in them mixed with Vegas and I doubt if this probem got in the way at all.

Bob.

I'm imagining a soundtrack in SONAR with audio sync'd precisely [i.e. exactly to the start of a frame, not just somewhere within it] to video cuts and transitions. If I understand what you are saying, the reference video doesn't need to be sub-frame-accurate because it would be pointless to attempt such precision composing.

So simply matching the overall project length [by stretching/shrinking as you suggested] will suffice.

This precision paranoia is probably a hold-over from all the sample-accurate audio editing I've done. Working in divisions of 24 or 30 per second feels sloppy by comparison to 44.1k/48k audio sample rates.

Thanks for the reality check.

The reference video can be frame accurate, it just doesn't need to be full raster perfect video, DV should be more than adequate, even if the footage was shot on 35mm. Composers want to see the action and the story. Rendering at anything from Draft to Best will not change frame accuracy, just image quality. Once you're done you will not use the reference video so the quality doesn't matter.

One other tip I picked up when doing this kind of work, don't use Dropped Frame TC in NTSC. I don't work in NTSC so I don't fully comprehend why this advice was given.

Bob.

Interesting...I think I understand why. But I'll do some more research on drop frame vs. non drop frame. It's been some time since I studied those.

Just to test my theory: Were you told this in a film production or TV production environment?


As for the original question: As long as I can just stretch/shrink the audio exported from SONAR to match the Vegas project [as you suggested], then I'm golden. For some reason I was thinking that doing so would throw off the timing of audio events for the entire length of the project.

"Just to test my theory: Were you told this in a film production or TV production environment?"

It was mentioned in this forum some time ago by one of the Vegas engineers. I think the issue might be that Vegas doesn't actually read TC from the file, it calcs it from the start TC. I've seen this work to my advantage when I had a tape with very wobbly TC. But with DF TC Vegas would have no way of knowing where the original TC dropped.

Might be better of someone from NTSC land chimed in on this one, thinking about it hurts my head and hopefully I'll never need to wrangle the DF/NDF problem myself.

Bob.